pdf

package module
v1.1.2 Latest Latest
Warning

This package is not in the latest version of its module.

 Go to latest Published: Nov 25, 2025 License: BSD-3-Clause Imports: 34 Imported by: 0

README

GoPDF - High-Performance PDF Processing Library

Go Version License Test Coverage

GoPDF is a powerful PDF processing library written in Go, focused on efficient text extraction, content analysis, and multilingual support. Built with a modular architecture, it provides high-performance concurrent processing capabilities.

✨ Key Features

📖 Text Extraction & Analysis
  • Intelligent Text Extraction: Supports plain text and styled text extraction
  • Semantic Classification: Automatic identification of titles, paragraphs, lists, tables, and other content types
  • Multilingual Support: Built-in English, French, German, and Spanish language detection and processing
  • Layout Analysis: Smart handling of multi-column layouts and complex page structures
🚀 Performance Optimization
  • Memory Optimization (NEW): Targeted allocation reduction for high-volume processing
    • Pre-allocated slices with capacity estimation (30-40% allocation reduction)
    • Eliminated unnecessary copies in hot paths (50% memory reduction in sorting)
    • Precise capacity calculation in merge operations (100+ allocations → 3)
    • Optimized string builder growth (40-50% reduction in string operations)
  • Sharded Caching: 256-shard cache with lock-free statistics (70-80% lock contention reduction)
  • Font Prefetching: Intelligent pattern-based font preloading with priority queuing
  • Zero-Copy Strings: Unsafe pointer optimization reducing memory allocation by 30-50%
  • Pool Warmup: Startup memory pool pre-warming reducing first-access latency by 60-80%
  • Enhanced Parallel Processing: Adaptive worker pools with batch processing (50% scheduling overhead reduction)
  • Memory Management: Advanced object pooling and resource management
  • Spatial Indexing: R-tree spatial indexing for optimized layout analysis
  • Asynchronous I/O: Streaming support for large files
🔧 Technical Features
  • Encoding Support: UTF-16, PDFDocEncoding, WinAnsi, MacRoman, and more
  • Compression Formats: Flate, LZW, ASCII85, RunLength
  • Encryption Support: RC4, AES encrypted PDFs
  • Thread Safety: Fully concurrent-safe operations
  • Error Handling: Detailed error context and recovery mechanisms

📦 Installation

go get -u github.com/Geek0x0/gopdf

🚀 Quick Start

Basic Text Extraction
package main

import (
    "fmt"
    "log"
    "github.com/Geek0x0/gopdf"
)

func main() {
    // Open PDF file
    file, reader, err := gopdf.Open("example.pdf")
    if err != nil {
        log.Fatal(err)
    }
    defer file.Close()

    // Extract plain text
    textReader, err := reader.GetPlainText()
    if err != nil {
        log.Fatal(err)
    }

    // Read text content
    // ... use textReader
}

⚡ Performance Quick Start

For high-performance PDF processing, follow these optimization steps:

1. Optimized Application Startup
import "github.com/Geek0x0/gopdf"

func init() {
    // Pre-warm memory pools and optimize GC settings
    config := pdf.DefaultStartupConfig()
    config.WarmupPools = true
    config.GCPercent = 200  // Reduce GC frequency
    
    if err := pdf.OptimizedStartup(config); err != nil {
        log.Fatalf("Startup optimization failed: %v", err)
    }
}
2. Use Parallel Extraction for Large Documents
func extractLargeDocument(filename string) ([]string, error) {
    f, r, err := pdf.Open(filename)
    if err != nil {
        return nil, err
    }
    defer f.Close()
    
    ctx, cancel := context.WithTimeout(context.Background(), 5*time.Minute)
    defer cancel()
    
    // Automatically uses all CPU cores
    return r.ExtractAllPagesParallel(ctx, 0)
}
3. Enable Caching for Repeated Operations
// Create global cache
var globalCache = pdf.NewShardedCache(100000, 1*time.Hour)

func getPageText(reader *pdf.Reader, pageNum int) (string, error) {
    cacheKey := fmt.Sprintf("page_%d", pageNum)
    
    // Check cache first
    if cached, ok := globalCache.Get(cacheKey); ok {
        return cached.(string), nil
    }
    
    // Extract and cache
    page := reader.Page(pageNum)
    text, err := page.GetPlainText(nil)
    if err == nil {
        globalCache.Set(cacheKey, text, int64(len(text)))
    }
    
    return text, err
}
4. Use Zero-Copy for String Operations
func processTexts(texts []string) string {
    // Fast zero-copy string operations
    builder := pdf.NewStringBuffer(10240)
    
    for _, text := range texts {
        trimmed := pdf.TrimSpaceZeroCopy(text)
        builder.WriteString(trimmed)
        builder.WriteByte('\n')
    }
    
    return builder.StringCopy()
}

🚀 Quick Start

Basic Text Extraction
package main

import (
    "fmt"
    "log"
    "github.com/Geek0x0/gopdf"
)

func main() {
    // Open PDF file
    file, reader, err := gopdf.Open("example.pdf")
    if err != nil {
        log.Fatal(err)
    }
    defer file.Close()

    // Extract plain text
    textReader, err := reader.GetPlainText()
    if err != nil {
        log.Fatal(err)
    }

    // Read text content
    // ... use textReader
}

// Extract text with context
textReader, err := reader.ExtractWithContext(ctx, gopdf.ExtractOptions{
    Workers: 4,  // Number of parallel worker threads
})

// Classify text blocks
blocks, err := page.ClassifyTextBlocks()
if err != nil {
    log.Fatal(err)
}

for _, block := range blocks {
    fmt.Printf("Type: %s, Text: %s\n", block.Type, block.Text)
}
High-Performance Parallel Extraction
import "context"

// Extract all pages in parallel with all optimizations
ctx, cancel := context.WithTimeout(context.Background(), 1*time.Minute)
defer cancel()

// Automatically uses runtime.NumCPU() workers when workers=0
pages, err := reader.ExtractAllPagesParallel(ctx, 0)
if err != nil {
    log.Fatal(err)
}

for i, text := range pages {
    fmt.Printf("Page %d: %d characters\n", i+1, len(text))
}
Using ParallelExtractor Directly
// Create parallel extractor with custom worker count
extractor := pdf.NewParallelExtractor(4)
defer extractor.Close()

// Collect pages
numPages := reader.NumPage()
pages := make([]pdf.Page, numPages)
for i := 0; i < numPages; i++ {
    pages[i] = reader.Page(i + 1)
    pages[i].SetFontCacheInterface(extractor.GetCache())
}

// Extract with context
results, err := extractor.ExtractAllPages(ctx, pages)
if err != nil {
    log.Fatal(err)
}

// Get performance stats
cacheStats := extractor.GetCacheStats()
fmt.Printf("Cache hits: %d, misses: %d\n", cacheStats.Hits, cacheStats.Misses)
Multilingual Text Processing
// Create multilingual processor
processor := gopdf.NewMultiLangProcessor()

// Detect text language
result := processor.DetectLanguage("Hello world! Bonjour le monde!")
fmt.Printf("Detected language: %s (confidence: %.2f)\n", result.Language, result.Confidence)

// Extract text by language
extractor := gopdf.NewLanguageTextExtractor()
textsByLang, err := extractor.ExtractTextByLanguage(reader)
Performance Optimization Features
// 1. Optimized Startup with Pool Warmup
err := pdf.OptimizedStartup(pdf.DefaultStartupConfig())
if err != nil {
    log.Fatal(err)
}

// 2. Sharded Cache (256 shards, lock-free)
cache := pdf.NewShardedCache(10000, 30*time.Minute)
cache.Set("key", value, 100)
if val, ok := cache.Get("key"); ok {
    // Use cached value
}
stats := cache.GetStats()
fmt.Printf("Hits: %d, Misses: %d, Evictions: %d\n", 
    stats.Hits, stats.Misses, stats.Evictions)

// 3. Font Prefetching (intelligent pattern-based)
fontCache := pdf.NewOptimizedFontCache(1000)
prefetcher := pdf.NewFontPrefetcher(fontCache)
defer prefetcher.Close()
prefetcher.RecordAccess("Arial", []string{"Helvetica", "Times"})

// 4. Zero-Copy String Operations
builder := pdf.NewStringBuffer(1024)
builder.WriteString("Hello")
builder.WriteByte(' ')
builder.WriteString("World")
result := builder.StringCopy()  // Safe copy

// Fast string operations
trimmed := pdf.TrimSpaceZeroCopy("  text  ")
parts := pdf.SplitZeroCopy("a,b,c", ',')
joined := pdf.JoinZeroCopy([]string{"a", "b", "c"}, ",")

🏗️ Architecture

GoPDF uses a modular architecture with clear component responsibilities:

gopdf/
├── read.go                      # PDF file reading and parsing
├── text.go                      # Core text extraction logic
├── page.go                      # Page structure analysis
├── metadata.go                  # Metadata processing
├── caching.go                   # Caching strategy implementation
├── spatial_index.go             # Spatial indexing (R-tree)
├── text_classifier.go           # Text classifier
├── multilang.go                 # Multilingual support
├── parallel_processing.go       # Parallel processing
├── performance.go               # Performance optimization
├── async_io.go                  # Asynchronous I/O
├── errors.go                    # Error handling
│
├── Performance Optimizations (2024)
├── sharded_cache.go             # 256-shard high-performance cache
├── font_prefetch.go             # Intelligent font prefetching
├── zero_copy_strings.go         # Zero-copy string operations
├── pool_warmup.go               # Memory pool pre-warming
├── enhanced_parallel.go         # Enhanced parallel processing
└── optimizations_advanced.go    # Advanced optimizations
Core Components
  • Reader: Main PDF reading interface with encryption support
  • Text Extractor: Intelligent text extraction engine with smart ordering
  • Classifier: ML-based text classification for semantic analysis
  • Sharded Cache: 256-shard lock-free cache system
  • Font Prefetcher: Pattern-based predictive font loading
  • Parallel Extractor: Adaptive worker pool with batch processing
  • Spatial Index: R-tree spatial query optimization
  • Language Processor: Multilingual detection and processing
  • Zero-Copy Optimizer: Memory allocation reduction utilities

📊 Performance Benchmarks

Performance metrics based on standard test datasets (Intel i7-14700K):

Overall Performance
  • Text Extraction Speed: Average 50-100 pages/second
  • Memory Usage: Smart object pooling, 40% reduction in memory footprint
  • Concurrent Processing: Multi-core support, 3-5x performance improvement with parallel extractor
Optimization Benchmarks
Sharded Cache Performance
  • Set Operations: ~114 ns/op (256 shards)
  • Get Operations: ~112 ns/op
  • Concurrent Access: ~32 ns/op (70-80% lock contention reduction)
  • Cache Hit Rate: Up to 85% with LRU policies
Zero-Copy String Operations
  • BytesToString: 0.12 ns/op (109x faster than standard)
  • String Concat: 10.44 ns/op (3.1x faster)
  • TrimSpace: 2.66 ns/op (1.2x faster)
  • Split: 60.51 ns/op (1.4x faster)
Memory Pool Warmup
  • Light Warmup: ~37 µs (development)
  • Default Warmup: ~96 µs (production)
  • Aggressive Warmup: ~358 µs (high-performance)
  • Concurrent vs Sequential: 35% faster with concurrent warmup
Parallel Extraction
  • 2 Workers: 1.8x speedup
  • 4 Workers: 3.1x speedup
  • 8 Workers: 5.0x speedup
  • Auto (CPU cores): 4.2x average speedup

🧪 Testing

The project maintains testing standards with 39.5% coverage (main package):

  • Unit tests covering all core functionality
  • Integration tests for end-to-end PDF processing
  • Performance tests with benchmarks and memory profiling
  • Concurrency tests for thread safety validation
  • Optimization-specific tests for new features
# Run all tests
go test ./...

# Run coverage tests
go test -coverprofile=coverage.out ./...
go tool cover -html=coverage.out

# Run performance benchmarks
go test -bench=. -benchmem -benchtime=500ms

# Run specific optimization benchmarks
go test -bench=BenchmarkShardedCache -run=^$
go test -bench=BenchmarkStringOperations -run=^$
go test -bench=BenchmarkParallelExtractor -run=^$
go test -bench=BenchmarkWarmup -run=^$
Benchmark Examples
# Compare parallel vs sequential extraction
go test -bench=BenchmarkParallelExtractorVsSequential -run=^$ -benchtime=500ms

# Zero-copy string operations performance
go test -bench=BenchmarkStringOperations -run=^$ -benchtime=500ms

# Cache performance under different loads
go test -bench=BenchmarkShardedCache -run=^$ -benchtime=1s

📚 API Documentation

Core Interfaces
// PDF file operations
Open(filename string) (*os.File, *Reader, error)
NewReader(r io.ReaderAt, size int64) (*Reader, error)

// Text extraction
(reader *Reader) GetPlainText() (io.Reader, error)
(reader *Reader) ExtractWithContext(ctx context.Context, opts ExtractOptions) (io.Reader, error)
(reader *Reader) ExtractAllPagesParallel(ctx context.Context, workers int) ([]string, error)

// Page operations
(reader *Reader) Page(num int) *Page
(page *Page) Content() *Content
(page *Page) ClassifyTextBlocks() ([]ClassifiedBlock, error)

// High-Performance Parallel Extraction
NewParallelExtractor(workers int) *ParallelExtractor
(pe *ParallelExtractor) ExtractAllPages(ctx context.Context, pages []Page) ([][]Text, error)
(pe *ParallelExtractor) GetCacheStats() ShardedCacheStats
(pe *ParallelExtractor) GetPrefetchStats() PrefetchStats
(pe *ParallelExtractor) Close()

// Sharded Cache
NewShardedCache(maxSize int, ttl time.Duration) *ShardedCache
(sc *ShardedCache) Get(key string) (interface{}, bool)
(sc *ShardedCache) Set(key string, value interface{}, size int64)
(sc *ShardedCache) GetStats() ShardedCacheStats
(sc *ShardedCache) Clear()

// Font Prefetching
NewFontPrefetcher(cache *OptimizedFontCache) *FontPrefetcher
(fp *FontPrefetcher) RecordAccess(fontKey string, relatedKeys []string)
(fp *FontPrefetcher) GetStats() PrefetchStats
(fp *FontPrefetcher) Close()

// Zero-Copy String Operations
BytesToString(b []byte) string
StringToBytes(s string) []byte
NewStringBuffer(capacity int) *StringBuffer
FastStringConcatZC(parts ...string) string
TrimSpaceZeroCopy(s string) string
SplitZeroCopy(s string, sep byte) []string
JoinZeroCopy(parts []string, sep string) string

// Pool Warmup
OptimizedStartup(config *StartupConfig) error
WarmupGlobal(config *WarmupConfig) error
DefaultWarmupConfig() *WarmupConfig
Performance Optimization APIs
// Optimized startup (recommended at application start)
config := pdf.DefaultStartupConfig()
config.WarmupPools = true
config.PreallocateCaches = true
err := pdf.OptimizedStartup(config)

// Create optimized font cache
fontCache := pdf.NewOptimizedFontCache(1000)

// Use string pool for repeated strings
pool := pdf.NewStringPool()
fontName := pool.Intern("Arial")

🤝 Contributing

Contributions are welcome! Please follow these steps:

  1. Fork the repository
  2. Create a feature branch (git checkout -b feature/AmazingFeature)
  3. Commit your changes (git commit -m 'Add some AmazingFeature')
  4. Push to the branch (git push origin feature/AmazingFeature)
  5. Create a Pull Request
Development Setup
# Clone repository
git clone https://github.com/Geek0x0/gopdf.git
cd gopdf

# Install dependencies
go mod download

# Run tests
go test ./...

# Build examples
go build ./examples/...

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

🙏 Acknowledgments

  • Based on unidoc/unipdf PDF parsing technology
  • Valuable feedback and suggestions from community contributors
  • Excellent language and toolchain provided by the Go team

📞 Contact

⭐ If this project helps you, please give us a star!

Documentation

Overview

Package pdf implements reading of PDF files.

Overview

PDF is Adobe's Portable Document Format, ubiquitous on the internet. A PDF document is a complex data format built on a fairly simple structure. This package exposes the simple structure along with some wrappers to extract basic information. If more complex information is needed, it is possible to extract that information by interpreting the structure exposed by this package.

Specifically, a PDF is a data structure built from Values, each of which has one of the following Kinds: 

Null, for the null object.
Integer, for an integer.
Real, for a floating-point number.
Bool, for a boolean value.
Name, for a name constant (as in /Helvetica).
String, for a string constant.
Dict, for a dictionary of name-value pairs.
Array, for an array of values.
Stream, for an opaque data stream and associated header dictionary.

The accessors on Value—Int64, Float64, Bool, Name, and so on—return a view of the data as the given type. When there is no appropriate view, the accessor returns a zero result. For example, the Name accessor returns the empty string if called on a Value v for which v.Kind() != Name. Returning zero values this way, especially from the Dict and Array accessors, which themselves return Values, makes it possible to traverse a PDF quickly without writing any error checking. On the other hand, it means that mistakes can go unreported.

The basic structure of the PDF file is exposed as the graph of Values.

Most richer data structures in a PDF file are dictionaries with specific interpretations of the name-value pairs. The Font and Page wrappers make the interpretation of a specific Value as the corresponding type easier. They are only helpers, though: they are implemented only in terms of the Value API and could be moved outside the package. Equally important, traversal of other PDF data structures can be implemented in other packages as needed.

Example (ZeroCopyInPDFProcessing)

演示在实际 PDF 处理中如何使用零拷贝优化 

// 假设从 PDF 提取了一些文本块
texts := []string{
	"  First paragraph  ",
	"  Second paragraph  ",
	"  Third paragraph  ",
}

// 使用零拷贝操作处理
builder := NewStringBuffer(1024)

for i, text := range texts {
	// 去除首尾空格（零拷贝）
	trimmed := TrimSpaceZeroCopy(text)
	builder.WriteString(trimmed)

	if i < len(texts)-1 {
		builder.WriteString("\n")
	}
}

result := builder.StringCopy()
fmt.Println(result)

Output:

First paragraph
Second paragraph
Third paragraph

Index

Examples

Constants

This section is empty.

Variables

View Source 
var (
	// ErrInvalidFont indicates a font definition is malformed or unsupported
	ErrInvalidFont = errors.New("invalid or unsupported font")

	// ErrUnsupportedEncoding indicates the character encoding is not supported
	ErrUnsupportedEncoding = errors.New("unsupported character encoding")

	// ErrMalformedStream indicates a content stream is malformed
	ErrMalformedStream = errors.New("malformed content stream")

	// ErrInvalidPage indicates an invalid page number or corrupted page
	ErrInvalidPage = errors.New("invalid page")

	// ErrEncrypted indicates the PDF is encrypted and cannot be read without a password
	ErrEncrypted = errors.New("PDF is encrypted")

	// ErrCorrupted indicates the PDF file structure is corrupted
	ErrCorrupted = errors.New("PDF file is corrupted")

	// ErrUnsupportedVersion indicates the PDF version is not supported
	ErrUnsupportedVersion = errors.New("unsupported PDF version")

	// ErrNoContent indicates the page has no content
	ErrNoContent = errors.New("page has no content")
)

Common errors

View Source 
var DebugOn = false

DebugOn is responsible for logging messages into stdout. If problems arise during reading, set it true.

View Source 
var ErrInvalidPassword = fmt.Errorf("encrypted PDF: invalid password")
View Source 
var ErrMemoryLimitExceeded = errors.New("pdf: stream processor memory limit exceeded")
View Source 
var GlobalMetrics = &PerformanceMetrics{}

全局性能指标实例

View Source 
var GlobalPoolWarmer = &PoolWarmer{
	bytePool: globalSizedBytePool,
	textPool: globalSizedTextSlicePool,
}

GlobalPoolWarmer 全局池预热器实例

Functions

func AutoWarmup added in v1.0.2 

func AutoWarmup() error

AutoWarmup 自动预热（根据可用内存选择配置）

func BatchCompareFloat64 

func BatchCompareFloat64(a, b []float64, threshold float64) []bool

9. SIMD友好的批量操作（伪代码，实际需要汇编）

func BenchmarkSortingAlgorithms added in v1.0.1 

func BenchmarkSortingAlgorithms(texts []Text, getCoord func(Text) float64) map[string]float64

BenchmarkSortingAlgorithms compares performance of different algorithms

func BytesToString added in v1.0.2 

func BytesToString(b []byte) string

BytesToString 零拷贝转换 []byte 到 string 警告：返回的字符串直接引用底层字节数组，不要修改原始 []byte

func ClearGlobalStringPool added in v1.0.2 

func ClearGlobalStringPool()

ClearGlobalStringPool 清空全局字符串池

func CompareStringsZeroCopy added in v1.0.2 

func CompareStringsZeroCopy(s1, s2 string) int

CompareStringsZeroCopy 零拷贝字符串比较 返回 -1 (s1 < s2), 0 (s1 == s2), 1 (s1 > s2)

func EstimateCapacity 

func EstimateCapacity(currentLen int, growthFactor float64) int

EstimateCapacity provides better capacity estimation for slices

func ExampleOptimizations 

func ExampleOptimizations()

使用示例

func FastSortTexts added in v1.0.1 

func FastSortTexts(texts []Text, less func(i, j int) bool)

FastSortTexts sorts texts using the fastest algorithm for the comparison function

func FastSortTextsByX added in v1.0.1 

func FastSortTextsByX(texts []Text)

FastSortTextsByX sorts texts by X coordinate using the fastest algorithm

func FastSortTextsByY added in v1.0.1 

func FastSortTextsByY(texts []Text)

FastSortTextsByY sorts texts by Y coordinate using the fastest algorithm

func FastStringConcat 

func FastStringConcat(strings ...string) string

FastStringConcat concatenates strings with optimized memory allocation

func FastStringConcatZC added in v1.0.2 

func FastStringConcatZC(parts ...string) string

FastStringConcatZC 快速拼接多个字符串（零拷贝版本）

Example

ExampleFastStringConcatZC 演示快速字符串拼接 

result := FastStringConcatZC("Hello", " ", "World", "!")
fmt.Println(result)

Output:

Hello World!

func FastStringSearch 

func FastStringSearch(haystack, needle string) int

FastStringSearch performs optimized string search using SIMD-like operations This is a simplified implementation that can be extended with actual SIMD instructions

func GetBuilder 

func GetBuilder() *strings.Builder

GetBuilder retrieves a strings.Builder from the pool

func GetByteBuffer 

func GetByteBuffer() *[]byte

GetByteBuffer retrieves a byte buffer from the pool

func GetPDFBuffer 

func GetPDFBuffer() *buffer

GetPDFBuffer retrieves a PDF buffer from the pool

func GetSizedBuffer added in v1.0.1 

func GetSizedBuffer(size int) []byte

GetSizedBuffer retrieves a byte buffer from the global sized pool This is a convenience function for common use cases

func HasPrefixZeroCopy added in v1.0.2 

func HasPrefixZeroCopy(s, prefix string) bool

HasPrefixZeroCopy 零拷贝前缀检查

func HasSuffixZeroCopy added in v1.0.2 

func HasSuffixZeroCopy(s, suffix string) bool

HasSuffixZeroCopy 零拷贝后缀检查

func HilbertXYToIndex 

func HilbertXYToIndex(x, y, order uint32) uint64

8. Hilbert曲线计算（用于空间索引）

func InternString added in v1.0.2 

func InternString(s string) string

InternString 将字符串加入全局池

func Interpret 

func Interpret(strm Value, do func(stk *Stack, op string))

Interpret interprets the content in a stream as a basic PostScript program, pushing values onto a stack and then calling the do function to execute operators. The do function may push or pop values from the stack as needed to implement op.

Interpret handles the operators "dict", "currentdict", "begin", "end", "def", and "pop" itself.

Interpret is not a full-blown PostScript interpreter. Its job is to handle the very limited PostScript found in certain supporting file formats embedded in PDF files, such as cmap files that describe the mapping from font code points to Unicode code points.

A stream can also be represented by an array of streams that has to be handled as a single stream In the case of a simple stream read only once, otherwise get the length of the stream to handle it properly

There is no support for executable blocks, among other limitations.

func IsSameSentence 

func IsSameSentence(last, current Text) bool

isSameSentence checks if the current text segment likely belongs to the same sentence as the last text segment based on font, size, vertical position, and lack of sentence-ending punctuation in the last segment.

func JoinZeroCopy added in v1.0.2 

func JoinZeroCopy(parts []string, sep string) string

JoinZeroCopy 零拷贝字符串连接（一次性分配）

Example

ExampleJoinZeroCopy 演示零拷贝连接 

parts := []string{"apple", "banana", "cherry"}
result := JoinZeroCopy(parts, ", ")
fmt.Println(result)

Output:

apple, banana, cherry

func OptimizedStartup added in v1.0.2 

func OptimizedStartup(config *StartupConfig) error

OptimizedStartup 优化启动流程 包括池预热、缓存预分配等

func PreallocateCache added in v1.0.2 

func PreallocateCache(fontCacheSize, resultCacheSize int)

PreallocateCache 预分配缓存（附加功能）

func ProcessLargePDF 

func ProcessLargePDF(reader *Reader, chunkSize, bufferSize int, maxMemory int64,
	handler func(PageStream) error) error

ProcessLargePDF handles very large PDFs with streaming

func ProcessTextWithMultiLanguage 

func ProcessTextWithMultiLanguage(reader *Reader) (map[Language][]ClassifiedBlock, error)

ProcessTextWithMultiLanguage handles multi-language text processing for the entire PDF

func PutBlockSlice 

func PutBlockSlice(s []ClassifiedBlock)

PutBlockSlice returns a ClassifiedBlock slice to the pool

func PutBuilder 

func PutBuilder(b *strings.Builder)

PutBuilder returns a strings.Builder to the pool after resetting it

func PutByteBuffer 

func PutByteBuffer(buf *[]byte)

PutByteBuffer returns a byte buffer to the pool

func PutPDFBuffer 

func PutPDFBuffer(b *buffer)

PutPDFBuffer returns a PDF buffer to the pool after resetting

func PutSizedBuffer added in v1.0.1 

func PutSizedBuffer(buf []byte)

PutSizedBuffer returns a byte buffer to the global sized pool This is a convenience function for common use cases

func PutSizedStringBuilder added in v1.0.1 

func PutSizedStringBuilder(sb *FastStringBuilder, estimatedSize int)

PutSizedStringBuilder returns a string builder to the appropriate pool

func PutSizedTextSlice added in v1.0.1 

func PutSizedTextSlice(slice []Text)

PutSizedTextSlice returns a Text slice to the global pool

func PutText 

func PutText(t *Text)

PutText returns a Text object to the appropriate pool

func PutTextSlice 

func PutTextSlice(s []Text)

PutTextSlice returns a Text slice to the pool

func RadixSortFloat64 

func RadixSortFloat64(values []float64)

7. Radix Sort 浮点数优化

func ResetSortingMetrics added in v1.0.1 

func ResetSortingMetrics()

ResetSortingMetrics resets the sorting metrics

func SmartTextRunsToPlain 

func SmartTextRunsToPlain(texts []Text) string

SmartTextRunsToPlain converts text runs to plain text using improved ordering

func SplitZeroCopy added in v1.0.2 

func SplitZeroCopy(s string, sep byte) []string

SplitZeroCopy 零拷贝字符串分割 返回的切片中的字符串都是原字符串的切片

Example

ExampleSplitZeroCopy 演示零拷贝分割 

str := "a,b,c,d"
parts := SplitZeroCopy(str, ',')
for _, part := range parts {
	fmt.Println(part)
}

Output:

a
b
c
d

func StringSliceToByteSlice added in v1.0.2 

func StringSliceToByteSlice(strings []string) [][]byte

StringSliceToByteSlice 零拷贝转换 []string 中的每个字符串 返回的 [][]byte 中每个元素都是只读的

func StringToBytes added in v1.0.2 

func StringToBytes(s string) []byte

StringToBytes 零拷贝转换 string 到 []byte 警告：返回的 []byte 是只读的，不要修改

func SubstringZeroCopy added in v1.0.2 

func SubstringZeroCopy(s string, start, end int) string

SubstringZeroCopy 零拷贝获取子字符串 实际上所有字符串切片在 Go 中已经是零拷贝的

func TrimSpaceZeroCopy added in v1.0.2 

func TrimSpaceZeroCopy(s string) string

TrimSpaceZeroCopy 零拷贝去除首尾空格

Example

ExampleTrimSpaceZeroCopy 演示零拷贝去空格 

str := "   hello world   "
result := TrimSpaceZeroCopy(str)
fmt.Println(result)

Output:

hello world

func WarmupGlobal added in v1.0.2 

func WarmupGlobal(config *WarmupConfig) error

WarmupGlobal 预热全局内存池（便捷函数）

func ZeroCopyStringSlice 

func ZeroCopyStringSlice(data []byte, separators []byte) []string

ZeroCopyStringSlice creates a string slice without copying data WARNING: This is unsafe and the returned strings share memory with the input

Types

type AccessPattern added in v1.0.2 

type AccessPattern struct {
	// contains filtered or unexported fields
}

AccessPattern 记录单个字体的访问模式

type AccessPatternTracker added in v1.0.2 

type AccessPatternTracker struct {
	// contains filtered or unexported fields
}

AccessPatternTracker 跟踪字体访问模式

type AdaptiveCapacityEstimator 

type AdaptiveCapacityEstimator struct {
	// contains filtered or unexported fields
}

AdaptiveCapacityEstimator 自适应容量估算器 基于历史数据动态调整预分配容量，减少重新分配

func NewAdaptiveCapacityEstimator 

func NewAdaptiveCapacityEstimator(maxSamples int) *AdaptiveCapacityEstimator

NewAdaptiveCapacityEstimator 创建新的自适应估算器

func (*AdaptiveCapacityEstimator) Estimate 

func (ace *AdaptiveCapacityEstimator) Estimate(hint int) int

Estimate 基于历史数据估算所需容量

func (*AdaptiveCapacityEstimator) Record 

func (ace *AdaptiveCapacityEstimator) Record(actual int)

Record 记录实际使用的容量

type AdaptiveProcessor added in v1.0.2 

type AdaptiveProcessor struct {
	// contains filtered or unexported fields
}

AdaptiveProcessor 自适应处理器 根据系统负载自动调整并发级别

func NewAdaptiveProcessor added in v1.0.2 

func NewAdaptiveProcessor(min, max int) *AdaptiveProcessor

NewAdaptiveProcessor 创建自适应处理器

func (*AdaptiveProcessor) AdjustWorkers added in v1.0.2 

func (ap *AdaptiveProcessor) AdjustWorkers()

AdjustWorkers 根据系统负载调整工作数

func (*AdaptiveProcessor) GetWorkerCount added in v1.0.2 

func (ap *AdaptiveProcessor) GetWorkerCount() int

GetWorkerCount 获取当前工作协程数

func (*AdaptiveProcessor) ProcessAdaptive added in v1.0.2 

func (ap *AdaptiveProcessor) ProcessAdaptive(
	ctx context.Context,
	pages []Page,
	processorFunc func(Page) ([]Text, error),
) ([][]Text, error)

ProcessAdaptive 自适应处理

type AdaptiveSorter added in v1.0.1 

type AdaptiveSorter struct {
	// contains filtered or unexported fields
}

AdaptiveSorter selects the best sorting algorithm based on data characteristics

func NewAdaptiveSorter added in v1.0.1 

func NewAdaptiveSorter() *AdaptiveSorter

NewAdaptiveSorter creates a new adaptive sorter with default thresholds

func (*AdaptiveSorter) SortTextsByComparison added in v1.0.1 

func (as *AdaptiveSorter) SortTextsByComparison(texts []Text, less func(i, j int) bool)

SortTextsByComparison sorts texts using a comparison function

func (*AdaptiveSorter) SortTextsByCoordinate added in v1.0.1 

func (as *AdaptiveSorter) SortTextsByCoordinate(texts []Text, getCoord func(Text) float64)

SortTextsByCoordinate sorts texts by a numeric coordinate using the best algorithm

type AsyncReader 

type AsyncReader struct {
	*Reader
	// contains filtered or unexported fields
}

AsyncReader wraps a Reader to provide asynchronous operations

func NewAsyncReader 

func NewAsyncReader(reader *Reader) *AsyncReader

NewAsyncReader creates a new async reader with async I/O support

func (*AsyncReader) AsyncExtractStructured 

func (ar *AsyncReader) AsyncExtractStructured(ctx context.Context) (<-chan []ClassifiedBlock, <-chan error)

AsyncExtractStructured extracts structured text asynchronously

func (*AsyncReader) AsyncExtractText 

func (ar *AsyncReader) AsyncExtractText(ctx context.Context) (<-chan string, <-chan error)

AsyncExtractText extracts text from all pages asynchronously

func (*AsyncReader) AsyncExtractTextWithContext 

func (ar *AsyncReader) AsyncExtractTextWithContext(ctx context.Context, opts ExtractOptions) (<-chan string, <-chan error)

AsyncExtractTextWithContext extracts text with cancellation and timeout support

func (*AsyncReader) AsyncStream 

func (ar *AsyncReader) AsyncStream(ctx context.Context, processor func(Page, int) error) <-chan error

AsyncStream processes the PDF file with async I/O operations

func (*AsyncReader) StreamValueReader 

func (ar *AsyncReader) StreamValueReader(ctx context.Context, v Value) (<-chan []byte, <-chan error)

StreamValueReader provides async streaming of value data

type AsyncReaderAt 

type AsyncReaderAt struct {
	// contains filtered or unexported fields
}

AsyncReaderAt provides async I/O for low-level file operations

func NewAsyncReaderAt 

func NewAsyncReaderAt(reader io.ReaderAt) *AsyncReaderAt

NewAsyncReaderAt creates a new async reader with async I/O support

func (*AsyncReaderAt) ReadAtAsync 

func (ara *AsyncReaderAt) ReadAtAsync(ctx context.Context, buf []byte, offset int64) (<-chan int, <-chan error)

ReadAtAsync reads from the file asynchronously

type BatchExtractOptions added in v1.0.1 

type BatchExtractOptions struct {
	// Pages to extract (nil means all pages)
	Pages []int

	// Number of concurrent workers (0 = NumCPU)
	Workers int

	// Whether to use smart text ordering
	SmartOrdering bool

	// Context for cancellation
	Context context.Context

	// Buffer size for each page result (0 = default 2KB)
	PageBufferSize int

	// Whether to enable font cache for this batch (default: false)
	// When enabled, a temporary font cache is created for the batch
	// to reduce redundant font parsing across pages
	UseFontCache bool

	// Maximum number of fonts to cache (0 = default 1000)
	// Only used when UseFontCache is true
	FontCacheSize int

	// FontCacheType specifies which cache implementation to use
	// - FontCacheStandard: Standard implementation (default)
	// - FontCacheOptimized: High-performance optimized cache (10-85x faster)
	// Only used when UseFontCache is true
	FontCacheType FontCacheType
}

BatchExtractOptions configures batch extraction behavior

Example (OptimizedCache)

ExampleBatchExtractOptions_optimizedCache demonstrates using optimized cache 

// This example shows how to use the optimized cache
opts := BatchExtractOptions{
	Workers:       8,
	SmartOrdering: true,
	UseFontCache:  true,
	FontCacheType: FontCacheOptimized, // Optimized cache (10-85x faster)
	FontCacheSize: 2000,
}

fmt.Printf("Cache type: Optimized, Size: %d\n", opts.FontCacheSize)

Output:

Cache type: Optimized, Size: 2000
Example (StandardCache)

ExampleBatchExtractOptions_standardCache demonstrates using standard cache 

// This example shows how to use the standard cache
opts := BatchExtractOptions{
	Workers:       4,
	SmartOrdering: true,
	UseFontCache:  true,
	FontCacheType: FontCacheStandard, // Standard cache
	FontCacheSize: 1000,
}

fmt.Printf("Cache type: Standard, Size: %d\n", opts.FontCacheSize)

Output:

Cache type: Standard, Size: 1000

type BatchResult added in v1.0.1 

type BatchResult struct {
	PageNum int
	Text    string
	Error   error
}

BatchResult contains the result of extracting a single page

type BatchStringBuilder 

type BatchStringBuilder struct {
	// contains filtered or unexported fields
}

BatchStringBuilder 批量字符串构建器 通过精确计算所需容量，避免多次重新分配

func NewBatchStringBuilder 

func NewBatchStringBuilder(texts []Text) *BatchStringBuilder

NewBatchStringBuilder 创建批量字符串构建器

func (*BatchStringBuilder) AppendTexts 

func (bsb *BatchStringBuilder) AppendTexts(texts []Text) string

AppendTexts 批量追加文本内容

func (*BatchStringBuilder) Reset 

func (bsb *BatchStringBuilder) Reset()

Reset 重置构建器以便重用

func (*BatchStringBuilder) String 

func (bsb *BatchStringBuilder) String() string

String 返回构建的字符串

type BlockType 

type BlockType int

BlockType represents the semantic type of a text block 

const (
	BlockUnknown   BlockType = iota
	BlockTitle               // Title or heading
	BlockParagraph           // Regular paragraph
	BlockList                // List item (numbered or bulleted)
	BlockCaption             // Image or table caption
	BlockFootnote            // Footnote or endnote
	BlockHeader              // Page header
	BlockFooter              // Page footer
)

func (BlockType) String 

func (bt BlockType) String() string

String returns the string representation of BlockType

type CacheContext 

type CacheContext struct {
	// contains filtered or unexported fields
}

CacheContext provides a context-aware cache with automatic cleanup

func NewCacheContext 

func NewCacheContext(parent context.Context, cache *ResultCache) *CacheContext

NewCacheContext creates a new context-aware cache

func (*CacheContext) Close 

func (cc *CacheContext) Close()

Close releases resources used by the cache context

func (*CacheContext) GetWithTimeout 

func (cc *CacheContext) GetWithTimeout(key string, timeout time.Duration) (interface{}, bool, error)

GetWithTimeout gets a value with timeout

type CacheEntry 

type CacheEntry struct {
	Data        interface{}
	Expiration  time.Time
	AccessCount int64
	LastAccess  time.Time
	Size        int64 // Estimated size in bytes
}

CacheEntry represents a cached item

func (*CacheEntry) IsExpired 

func (ce *CacheEntry) IsExpired() bool

IsExpired checks if the cache entry has expired

type CacheKeyGenerator 

type CacheKeyGenerator struct{}

CacheKeyGenerator provides functions to generate cache keys

func NewCacheKeyGenerator 

func NewCacheKeyGenerator() *CacheKeyGenerator

NewCacheKeyGenerator creates a new key generator

func (*CacheKeyGenerator) GenerateFullHash 

func (ckg *CacheKeyGenerator) GenerateFullHash(data string) string

GenerateFullHash generates a hash from arbitrary data

func (*CacheKeyGenerator) GeneratePageContentKey 

func (ckg *CacheKeyGenerator) GeneratePageContentKey(pageNum int, readerHash string) string

GeneratePageContentKey generates a cache key for page content

func (*CacheKeyGenerator) GenerateReaderHash 

func (ckg *CacheKeyGenerator) GenerateReaderHash(reader *Reader) string

GenerateReaderHash generates a hash for the reader object (simplified)

func (*CacheKeyGenerator) GenerateTextClassificationKey 

func (ckg *CacheKeyGenerator) GenerateTextClassificationKey(pageNum int, readerHash string, processorParams string) string

GenerateTextClassificationKey generates a cache key for text classification

func (*CacheKeyGenerator) GenerateTextOrderingKey 

func (ckg *CacheKeyGenerator) GenerateTextOrderingKey(pageNum int, readerHash string, orderingParams string) string

GenerateTextOrderingKey generates a cache key for text ordering

type CacheLineAlignedCounter 

type CacheLineAlignedCounter struct {
	// contains filtered or unexported fields
}

func NewCacheLineAlignedCounter 

func NewCacheLineAlignedCounter(n int) *CacheLineAlignedCounter

func (*CacheLineAlignedCounter) Add 

func (c *CacheLineAlignedCounter) Add(idx int, delta uint64)

func (*CacheLineAlignedCounter) Get 

func (c *CacheLineAlignedCounter) Get(idx int) uint64

type CacheLinePadded 

type CacheLinePadded struct {
	// contains filtered or unexported fields
}

4. 缓存行对齐结构

type CacheManager 

type CacheManager struct {
	// contains filtered or unexported fields
}

CacheManager provides centralized cache management

func NewCacheManager 

func NewCacheManager() *CacheManager

NewCacheManager creates a new cache manager with separate caches for different data types

func (*CacheManager) GetClassificationCache 

func (cm *CacheManager) GetClassificationCache() *ResultCache

GetClassificationCache returns the classification cache

func (*CacheManager) GetMetadataCache 

func (cm *CacheManager) GetMetadataCache() *ResultCache

GetMetadataCache returns the metadata cache

func (*CacheManager) GetPageCache 

func (cm *CacheManager) GetPageCache() *ResultCache

GetPageCache returns the page content cache

func (*CacheManager) GetTextOrderingCache 

func (cm *CacheManager) GetTextOrderingCache() *ResultCache

GetTextOrderingCache returns the text ordering cache

func (*CacheManager) GetTotalStats 

func (cm *CacheManager) GetTotalStats() CacheStats

GetTotalStats returns combined statistics for all caches

type CacheShard added in v1.0.2 

type CacheShard struct {
	// contains filtered or unexported fields
}

CacheShard 代表缓存的单个分片

type CacheStats 

type CacheStats struct {
	Hits        int64
	Misses      int64
	Evictions   int64
	CurrentSize int64
	MaxSize     int64
	Entries     int64
}

CacheStats provides statistics about cache performance

type CachedReader 

type CachedReader struct {
	*Reader
	// contains filtered or unexported fields
}

CachedReader wraps a Reader to provide caching functionality

func NewCachedReader 

func NewCachedReader(reader *Reader, cache *ResultCache) *CachedReader

NewCachedReader creates a new cached reader

func (*CachedReader) CachedClassifyTextBlocks 

func (cr *CachedReader) CachedClassifyTextBlocks(pageNum int) ([]ClassifiedBlock, error)

CachedClassifyTextBlocks returns classified text blocks with caching

func (*CachedReader) CachedPage 

func (cr *CachedReader) CachedPage(pageNum int) ([]Text, error)

CachedPage returns page content with caching

type ClassifiedBlock 

type ClassifiedBlock struct {
	Type    BlockType // Semantic type of the block
	Level   int       // Hierarchy level (for titles: 1=h1, 2=h2, etc.)
	Content []Text    // Text runs in this block
	Bounds  Rect      // Bounding box
	Text    string    // Concatenated text content
}

ClassifiedBlock represents a classified block of text with semantic information

func GetBlockSlice 

func GetBlockSlice() []ClassifiedBlock

GetBlockSlice retrieves a ClassifiedBlock slice from the pool

func GetTextByType 

func GetTextByType(blocks []ClassifiedBlock, blockType BlockType) []ClassifiedBlock

GetTextByType returns all text blocks of a specific type

func GetTitles 

func GetTitles(blocks []ClassifiedBlock, level int) []ClassifiedBlock

GetTitles returns all title blocks, optionally filtered by level

type ClassifiedBlockWithLanguage 

type ClassifiedBlockWithLanguage struct {
	ClassifiedBlock
	Language LanguageInfo
}

ClassifiedBlockWithLanguage represents a classified block with language information

type Column 

type Column struct {
	Position int64
	Content  TextVertical
}

Column represents the contents of a column

type Columns 

type Columns []*Column

Columns is a list of column

type ConnectionPool 

type ConnectionPool struct {
	// contains filtered or unexported fields
}

ConnectionPool manages a pool of connections/resources

func NewConnectionPool 

func NewConnectionPool(maxSize int, newFunc func() interface{}, closeFunc func(interface{})) *ConnectionPool

NewConnectionPool creates a new connection pool

func (*ConnectionPool) Close 

func (cp *ConnectionPool) Close()

Close closes all connections in the pool

func (*ConnectionPool) Get 

func (cp *ConnectionPool) Get() interface{}

Get retrieves a connection from the pool

func (*ConnectionPool) Put 

func (cp *ConnectionPool) Put(conn interface{})

Put returns a connection to the pool

type Content 

type Content struct {
	Text []Text
	Rect []Rect
}

Content describes the basic content on a page: the text and any drawn rectangles.

type EnhancedParallelProcessor added in v1.0.2 

type EnhancedParallelProcessor struct {
	// contains filtered or unexported fields
}

EnhancedParallelProcessor 增强的并行处理器 提供更好的并发控制、负载均衡和错误处理

func NewEnhancedParallelProcessor added in v1.0.2 

func NewEnhancedParallelProcessor(workers int, batchSize int) *EnhancedParallelProcessor

NewEnhancedParallelProcessor 创建增强的并行处理器

func (*EnhancedParallelProcessor) ProcessPagesEnhanced added in v1.0.2 

func (epp *EnhancedParallelProcessor) ProcessPagesEnhanced(
	ctx context.Context,
	pages []Page,
	processorFunc func(Page) ([]Text, error),
) ([][]Text, error)

ProcessPagesEnhanced 增强的并行页面处理

func (*EnhancedParallelProcessor) ProcessWithLoadBalancing added in v1.0.2 

func (epp *EnhancedParallelProcessor) ProcessWithLoadBalancing(
	ctx context.Context,
	pages []Page,
	processorFunc func(Page) ([]Text, error),
) ([][]Text, error)

ProcessWithLoadBalancing 带负载均衡的处理

func (*EnhancedParallelProcessor) ProcessWithPipeline added in v1.0.2 

func (epp *EnhancedParallelProcessor) ProcessWithPipeline(
	ctx context.Context,
	pages []Page,
	stages []func(Page, []Text) ([]Text, error),
) ([][]Text, error)

ProcessWithPipeline 管道式处理

type ExtractMode 

type ExtractMode int

ExtractMode specifies the type of extraction to perform 

const (
	ModePlain      ExtractMode = iota // Plain text extraction
	ModeStyled                        // Text with style information
	ModeStructured                    // Structured text with classification
)

type ExtractOptions 

type ExtractOptions struct {
	Workers   int   // Number of concurrent workers (0 = use NumCPU)
	PageRange []int // Specific pages to extract (nil = all pages)
}

ExtractOptions configures text extraction behavior

type ExtractResult 

type ExtractResult struct {
	Text             string            // Plain text (for ModePlain)
	StyledTexts      []Text            // Styled texts (for ModeStyled)
	ClassifiedBlocks []ClassifiedBlock // Classified blocks (for ModeStructured)
	Metadata         Metadata          // Document metadata
	PageCount        int               // Total number of pages
}

ExtractResult contains the results of text extraction

type Extractor 

type Extractor struct {
	// contains filtered or unexported fields
}

Extractor provides a builder pattern for configuring and executing extraction

func NewExtractor 

func NewExtractor(r *Reader) *Extractor

NewExtractor creates a new extractor for the given reader

func (*Extractor) Context 

func (e *Extractor) Context(ctx context.Context) *Extractor

Context sets the context for cancellation

func (*Extractor) Extract 

func (e *Extractor) Extract() (*ExtractResult, error)

Extract performs the extraction and returns the result

func (*Extractor) ExtractStructured 

func (e *Extractor) ExtractStructured() ([]ClassifiedBlock, error)

ExtractStructured is a convenience method for extracting structured text

func (*Extractor) ExtractStyledTexts 

func (e *Extractor) ExtractStyledTexts() ([]Text, error)

ExtractStyledTexts is a convenience method for extracting styled texts

func (*Extractor) ExtractText 

func (e *Extractor) ExtractText() (string, error)

ExtractText is a convenience method for extracting plain text

func (*Extractor) Mode 

func (e *Extractor) Mode(mode ExtractMode) *Extractor

Mode sets the extraction mode

func (*Extractor) Pages 

func (e *Extractor) Pages(pages ...int) *Extractor

Pages sets specific pages to extract (1-indexed)

func (*Extractor) SmartOrdering 

func (e *Extractor) SmartOrdering(enabled bool) *Extractor

SmartOrdering enables smart text ordering for multi-column layouts

func (*Extractor) Workers 

func (e *Extractor) Workers(n int) *Extractor

Workers sets the number of concurrent workers

type FastStringBuilder 

type FastStringBuilder struct {
	// contains filtered or unexported fields
}

FastStringBuilder provides optimized string building with pre-allocation

func GetSizedStringBuilder added in v1.0.1 

func GetSizedStringBuilder(estimatedSize int) *FastStringBuilder

GetSizedStringBuilder retrieves a string builder from the appropriate pool

func NewFastStringBuilder 

func NewFastStringBuilder(estimatedSize int) *FastStringBuilder

NewFastStringBuilder creates a builder with estimated capacity

func (*FastStringBuilder) Len 

func (b *FastStringBuilder) Len() int

Len returns the current length

func (*FastStringBuilder) Reset 

func (b *FastStringBuilder) Reset()

Reset clears the builder

func (*FastStringBuilder) String 

func (b *FastStringBuilder) String() string

String returns the accumulated string

func (*FastStringBuilder) WriteByte 

func (b *FastStringBuilder) WriteByte(c byte) error

WriteByte appends a byte

func (*FastStringBuilder) WriteString 

func (b *FastStringBuilder) WriteString(s string)

WriteString appends a string

type Font 

type Font struct {
	V Value
	// contains filtered or unexported fields
}

A Font represent a font in a PDF file. The methods interpret a Font dictionary stored in V.

func (Font) BaseFont 

func (f Font) BaseFont() string

BaseFont returns the font's name (BaseFont property).

func (*Font) Encoder 

func (f *Font) Encoder() TextEncoding

Encoder returns the encoding between font code point sequences and UTF-8. Pointer receiver is required so the computed encoder is cached on the shared Font instance instead of a copy. The previous value-receiver implementation rebuilt the encoder for every call, causing large allocations to pile up during batch extraction.

func (Font) FirstChar 

func (f Font) FirstChar() int

FirstChar returns the code point of the first character in the font.

func (Font) LastChar 

func (f Font) LastChar() int

LastChar returns the code point of the last character in the font.

func (Font) Width 

func (f Font) Width(code int) float64

Width returns the width of the given code point.

func (Font) Widths 

func (f Font) Widths() []float64

Widths returns the widths of the glyphs in the font. In a well-formed PDF, len(f.Widths()) == f.LastChar()+1 - f.FirstChar().

type FontCache 

type FontCache struct {
	// contains filtered or unexported fields
}

FontCache stores parsed fonts to avoid re-parsing across pages

func NewFontCache 

func NewFontCache() *FontCache

NewFontCache creates a new font cache

func (*FontCache) Get 

func (fc *FontCache) Get(key string) (*Font, bool)

Get retrieves a font from the cache

func (*FontCache) Set 

func (fc *FontCache) Set(key string, font *Font)

Set stores a font in the cache

type FontCacheInterface added in v1.0.1 

type FontCacheInterface interface {
	Get(key string) (*Font, bool)
	Set(key string, font *Font)
	Clear()
	GetStats() FontCacheStats
}

FontCacheInterface defines the common interface for font caches

type FontCacheStats added in v1.0.1 

type FontCacheStats struct {
	Entries     int
	MaxEntries  int
	Hits        uint64
	Misses      uint64
	HitRate     float64
	AvgAccesses float64
}

Stats returns cache statistics

type FontCacheType added in v1.0.1 

type FontCacheType int

FontCacheType specifies which font cache implementation to use 

const (
	// FontCacheStandard uses the standard GlobalFontCache (default)
	// - Stable and well-tested
	// - Good performance for most use cases
	// - Simpler implementation
	FontCacheStandard FontCacheType = iota

	// FontCacheOptimized uses the OptimizedFontCache
	// - 10-85x faster than standard (depending on workload)
	// - Lock-free read path with 16 shards
	// - Best for high-concurrency scenarios (>1000 qps)
	// - Recommended for production environments with heavy load
	FontCacheOptimized
)

type FontPrefetcher added in v1.0.2 

type FontPrefetcher struct {
	// contains filtered or unexported fields
}

FontPrefetcher 实现智能字体预取策略 基于访问模式预测和预加载可能需要的字体

func NewFontPrefetcher added in v1.0.2 

func NewFontPrefetcher(cache *OptimizedFontCache) *FontPrefetcher

NewFontPrefetcher 创建新的字体预取器

func (*FontPrefetcher) ClearPatterns added in v1.0.2 

func (fp *FontPrefetcher) ClearPatterns()

ClearPatterns 清除访问模式

func (*FontPrefetcher) Close added in v1.0.2 

func (fp *FontPrefetcher) Close()

Close 关闭预取器

func (*FontPrefetcher) Disable added in v1.0.2 

func (fp *FontPrefetcher) Disable()

Disable 禁用预取

func (*FontPrefetcher) Enable added in v1.0.2 

func (fp *FontPrefetcher) Enable()

Enable 启用预取

func (*FontPrefetcher) GetStats added in v1.0.2 

func (fp *FontPrefetcher) GetStats() PrefetchStats

GetStats 获取预取统计信息

func (*FontPrefetcher) RecordAccess added in v1.0.2 

func (fp *FontPrefetcher) RecordAccess(fontKey string, relatedKeys []string)

RecordAccess 记录字体访问

type GlobalFontCache added in v1.0.1 

type GlobalFontCache struct {
	// contains filtered or unexported fields
}

GlobalFontCache implements an enhanced global font cache with: - LRU eviction for memory control - Hit/miss statistics for monitoring - Content-based hashing for accurate cache keys

Example 
// Create a cache with max 100 entries and 1 hour expiration
cache := NewGlobalFontCache(100, 1*time.Hour)

// Store a font
font := &Font{}
cache.Set("MyFont", font)

// Retrieve the font
retrieved, ok := cache.Get("MyFont")
if ok {
	fmt.Println("Font found in cache")
	_ = retrieved
}

// Get statistics
stats := cache.GetStats()
fmt.Printf("Cache entries: %d, Hit rate: %.2f%%\n",
	stats.Entries, stats.HitRate*100)

func GetGlobalFontCache added in v1.0.1 

func GetGlobalFontCache() *GlobalFontCache

GetGlobalFontCache returns the global font cache instance

Example 
// Get the global singleton instance
cache := GetGlobalFontCache()

font := &Font{}
cache.Set("GlobalFont", font)

// The same instance can be accessed from anywhere
sameCacheInstance := GetGlobalFontCache()
retrieved, _ := sameCacheInstance.Get("GlobalFont")
_ = retrieved

func NewGlobalFontCache added in v1.0.1 

func NewGlobalFontCache(maxEntries int, maxAge time.Duration) *GlobalFontCache

NewGlobalFontCache creates a new global font cache

func (*GlobalFontCache) Cleanup added in v1.0.1 

func (gfc *GlobalFontCache) Cleanup() int

Cleanup removes expired entries

func (*GlobalFontCache) Clear added in v1.0.1 

func (gfc *GlobalFontCache) Clear()

Clear removes all fonts from the cache

func (*GlobalFontCache) Get added in v1.0.1 

func (gfc *GlobalFontCache) Get(key string) (*Font, bool)

Get retrieves a font from the cache

func (*GlobalFontCache) GetOrCompute added in v1.0.1 

func (gfc *GlobalFontCache) GetOrCompute(key string, compute func() (*Font, error)) (*Font, error)

GetOrCompute retrieves a font from cache or computes it if not present This is a convenience function that combines Get and Set

func (*GlobalFontCache) GetStats added in v1.0.1 

func (gfc *GlobalFontCache) GetStats() FontCacheStats

GetStats returns current cache statistics

func (*GlobalFontCache) Remove added in v1.0.1 

func (gfc *GlobalFontCache) Remove(key string)

Remove removes a font from the cache

func (*GlobalFontCache) Set added in v1.0.1 

func (gfc *GlobalFontCache) Set(key string, font *Font)

Set stores a font in the cache

func (*GlobalFontCache) StartCleanupRoutine added in v1.0.1 

func (gfc *GlobalFontCache) StartCleanupRoutine(interval time.Duration) chan struct{}

StartCleanupRoutine starts a background goroutine to periodically clean up expired entries

type GridKey 

type GridKey struct {
	X, Y int
}

GridKey represents a grid cell identifier

type InplaceStringBuilder added in v1.0.2 

type InplaceStringBuilder struct {
	// contains filtered or unexported fields
}

InplaceStringBuilder 原地字符串构建器 避免中间分配

func NewInplaceStringBuilder added in v1.0.2 

func NewInplaceStringBuilder(capacity int) *InplaceStringBuilder

NewInplaceStringBuilder 创建新的原地字符串构建器

func (*InplaceStringBuilder) Append added in v1.0.2 

func (isb *InplaceStringBuilder) Append(s string)

Append 追加字符串

func (*InplaceStringBuilder) Build added in v1.0.2 

func (isb *InplaceStringBuilder) Build() string

Build 构建最终字符串（一次性分配）

func (*InplaceStringBuilder) Len added in v1.0.2 

func (isb *InplaceStringBuilder) Len() int

Len 返回总长度

func (*InplaceStringBuilder) Reset added in v1.0.2 

func (isb *InplaceStringBuilder) Reset()

Reset 重置构建器

type KDNode 

type KDNode struct {
	// contains filtered or unexported fields
}

KDNode KD树节点

type KDTree 

type KDTree struct {
	// contains filtered or unexported fields
}

KDTree KD树空间索引 用于O(log n)时间复杂度的最近邻搜索

func BuildKDTree 

func BuildKDTree(blocks []*TextBlock) *KDTree

BuildKDTree 从文本块构建KD树

func (*KDTree) RangeSearch 

func (tree *KDTree) RangeSearch(target []float64, radius float64) []*TextBlock

RangeSearch 范围搜索，返回距离目标点在指定半径内的所有文本块 使用迭代方式替代递归,避免深度递归导致的大量内存分配

type Language 

type Language string

Language represents a language code 

const (
	English Language = "en"
	French  Language = "fr"
	German  Language = "de"
	Spanish Language = "es"
	Unknown Language = "unknown"
)

type LanguageInfo 

type LanguageInfo struct {
	Language      Language
	Confidence    float64 // Confidence level (0.0 to 1.0)
	Characters    []rune  // Unique characters in the text
	WordCount     int     // Number of words in the text
	SentenceCount int     // Number of sentences in the text
}

LanguageInfo contains information about a detected language

type LanguageTextExtractor 

type LanguageTextExtractor struct {
	// contains filtered or unexported fields
}

LanguageTextExtractor extracts text while detecting languages

func NewLanguageTextExtractor 

func NewLanguageTextExtractor() *LanguageTextExtractor

NewLanguageTextExtractor creates a new language-aware text extractor

func (*LanguageTextExtractor) ExtractTextByLanguage 

func (lte *LanguageTextExtractor) ExtractTextByLanguage(reader *Reader) (map[Language][]Text, error)

ExtractTextByLanguage extracts text grouped by detected language

func (*LanguageTextExtractor) GetLanguageStats 

func (lte *LanguageTextExtractor) GetLanguageStats(texts []Text) map[Language]int

GetLanguageStats returns statistics about languages detected in the text

func (*LanguageTextExtractor) GetTextsByLanguage 

func (lte *LanguageTextExtractor) GetTextsByLanguage(texts []Text, targetLang Language) []Text

GetTextsByLanguage returns text elements filtered by specific language

type LazyPage 

type LazyPage struct {
	// contains filtered or unexported fields
}

LazyPage provides lazy loading of page content to reduce memory usage for large PDFs where not all pages need to be processed

func NewLazyPage 

func NewLazyPage(r *Reader, pageNum int) *LazyPage

NewLazyPage creates a lazy-loading page wrapper

func (*LazyPage) GetContent 

func (lp *LazyPage) GetContent() *Content

GetContent loads and returns the page content (cached after first call)

func (*LazyPage) IsLoaded 

func (lp *LazyPage) IsLoaded() bool

IsLoaded returns whether the page content has been loaded

func (*LazyPage) Release 

func (lp *LazyPage) Release()

Release clears the cached content to free memory

type LazyPageManager 

type LazyPageManager struct {
	// contains filtered or unexported fields
}

LazyPageManager manages lazy loading of multiple pages

func NewLazyPageManager 

func NewLazyPageManager(r *Reader, maxCached int) *LazyPageManager

NewLazyPageManager creates a manager with LRU cache

func (*LazyPageManager) Clear 

func (m *LazyPageManager) Clear()

Clear releases all cached pages

func (*LazyPageManager) GetPage 

func (m *LazyPageManager) GetPage(pageNum int) *LazyPage

GetPage returns a lazy page, loading it if necessary

func (*LazyPageManager) GetStats 

func (m *LazyPageManager) GetStats() (totalPages, loadedPages int)

GetStats returns cache statistics

type LockFreeRingBuffer 

type LockFreeRingBuffer struct {
	// contains filtered or unexported fields
}

3. Lock-free 环形缓冲区（用于生产者-消费者）

func NewLockFreeRingBuffer 

func NewLockFreeRingBuffer(size int) *LockFreeRingBuffer

func (*LockFreeRingBuffer) Pop 

func (rb *LockFreeRingBuffer) Pop() (interface{}, bool)

func (*LockFreeRingBuffer) Push 

func (rb *LockFreeRingBuffer) Push(item interface{}) bool

type MemoryArena 

type MemoryArena struct {
	// contains filtered or unexported fields
}

10. 内存池管理器（减少GC压力）

func NewMemoryArena 

func NewMemoryArena(chunkSize int) *MemoryArena

func (*MemoryArena) Alloc 

func (a *MemoryArena) Alloc(size int) []byte

func (*MemoryArena) Reset 

func (a *MemoryArena) Reset()

type MemoryEfficientExtractor 

type MemoryEfficientExtractor struct {
	// contains filtered or unexported fields
}

MemoryEfficientExtractor provides memory-efficient extraction using streaming

func NewMemoryEfficientExtractor 

func NewMemoryEfficientExtractor(chunkSize, bufferSize int, maxMemory int64) *MemoryEfficientExtractor

NewMemoryEfficientExtractor creates a new memory-efficient extractor

func (*MemoryEfficientExtractor) ExtractTextStream 

func (mee *MemoryEfficientExtractor) ExtractTextStream(reader *Reader) (<-chan TextStream, <-chan error)

ExtractTextStream extracts text in a memory-efficient streaming way

func (*MemoryEfficientExtractor) ExtractTextToWriter 

func (mee *MemoryEfficientExtractor) ExtractTextToWriter(reader *Reader, writer io.Writer) (err error)

ExtractTextToWriter extracts text directly to an io.Writer to minimize memory usage

type Metadata 

type Metadata struct {
	Title        string            // Document title
	Author       string            // Author name
	Subject      string            // Document subject
	Keywords     []string          // Keywords
	Creator      string            // Application that created the document
	Producer     string            // PDF producer (converter)
	CreationDate time.Time         // Creation date
	ModDate      time.Time         // Last modification date
	Trapped      string            // Trapping information (True/False/Unknown)
	Custom       map[string]string // Custom metadata fields
}

Metadata represents PDF document metadata

func (Metadata) String 

func (m Metadata) String() string

GetDocumentInfo returns a formatted string with document information

type MultiLangProcessor 

type MultiLangProcessor struct {
	// contains filtered or unexported fields
}

MultiLangProcessor provides multi-language text processing

func NewMultiLangProcessor 

func NewMultiLangProcessor() *MultiLangProcessor

NewMultiLangProcessor creates a new multi-language processor

func (*MultiLangProcessor) DetectLanguage 

func (mlp *MultiLangProcessor) DetectLanguage(text string) LanguageInfo

DetectLanguage detects the language of a given text

func (*MultiLangProcessor) GetLanguageConfidenceThreshold 

func (mlp *MultiLangProcessor) GetLanguageConfidenceThreshold() float64

GetLanguageConfidenceThreshold returns a confidence threshold for reliable detection

func (*MultiLangProcessor) GetLanguageName 

func (mlp *MultiLangProcessor) GetLanguageName(lang Language) string

GetLanguageName returns the full name of a language

func (*MultiLangProcessor) GetSupportedLanguages 

func (mlp *MultiLangProcessor) GetSupportedLanguages() []Language

GetSupportedLanguages returns the list of supported languages

func (*MultiLangProcessor) IsEnglish 

func (mlp *MultiLangProcessor) IsEnglish(text string) bool

IsEnglish checks if text is likely English

func (*MultiLangProcessor) IsFrench 

func (mlp *MultiLangProcessor) IsFrench(text string) bool

IsFrench checks if text is likely French

func (*MultiLangProcessor) IsGerman 

func (mlp *MultiLangProcessor) IsGerman(text string) bool

IsGerman checks if text is likely German

func (*MultiLangProcessor) IsSpanish 

func (mlp *MultiLangProcessor) IsSpanish(text string) bool

IsSpanish checks if text is likely Spanish

func (*MultiLangProcessor) ProcessTextWithLanguageDetection 

func (mlp *MultiLangProcessor) ProcessTextWithLanguageDetection(texts []Text) []TextWithLanguage

ProcessTextWithLanguageDetection processes text with language detection

type MultiLanguageTextClassifier 

type MultiLanguageTextClassifier struct {
	*TextClassifier
	// contains filtered or unexported fields
}

MultiLanguageTextClassifier extends the text classifier with language awareness

func NewMultiLanguageTextClassifier 

func NewMultiLanguageTextClassifier(texts []Text, pageWidth, pageHeight float64) *MultiLanguageTextClassifier

NewMultiLanguageTextClassifier creates a new multi-language text classifier

func (*MultiLanguageTextClassifier) ClassifyBlocksWithLanguage 

func (mltc *MultiLanguageTextClassifier) ClassifyBlocksWithLanguage() []ClassifiedBlockWithLanguage

ClassifyBlocksWithLanguage extends the classification with language information

type MultiLevelCache 

type MultiLevelCache struct {
	// contains filtered or unexported fields
}

MultiLevelCache 多级缓存管理器

func NewMultiLevelCache 

func NewMultiLevelCache() *MultiLevelCache

NewMultiLevelCache 创建多级缓存

func (*MultiLevelCache) Get 

func (mlc *MultiLevelCache) Get(key string) (interface{}, bool)

Get 从缓存获取数据

func (*MultiLevelCache) Prefetch 

func (mlc *MultiLevelCache) Prefetch(keys []string)

Prefetch 预取页面数据

func (*MultiLevelCache) Put 

func (mlc *MultiLevelCache) Put(key string, value interface{})

Put 存入缓存

func (*MultiLevelCache) Stats 

func (mlc *MultiLevelCache) Stats() map[string]uint64

Stats 获取缓存统计

type OptimizedFontCache added in v1.0.1 

type OptimizedFontCache struct {
	// contains filtered or unexported fields
}

OptimizedFontCache implements an ultra-high-performance font cache with: - Lock-free read path using atomic operations - Sharded design to reduce lock contention (16 shards) - Zero-allocation fast path for cache hits - Inline LRU using lock-free linked list approximation - Pre-allocated pools for metadata structs - SIMD-friendly memory layout

func NewOptimizedFontCache added in v1.0.1 

func NewOptimizedFontCache(totalCapacity int) *OptimizedFontCache

NewOptimizedFontCache creates a new optimized font cache

func (*OptimizedFontCache) Clear added in v1.0.1 

func (ofc *OptimizedFontCache) Clear()

Clear removes all entries from all shards

func (*OptimizedFontCache) Get added in v1.0.1 

func (ofc *OptimizedFontCache) Get(key string) (*Font, bool)

Get retrieves a font from the cache (lock-free fast path)

func (*OptimizedFontCache) GetOrCompute added in v1.0.1 

func (ofc *OptimizedFontCache) GetOrCompute(key string, compute func() (*Font, error)) (*Font, error)

GetOrCompute retrieves a font from cache or computes it if not present

func (*OptimizedFontCache) GetStats added in v1.0.1 

func (ofc *OptimizedFontCache) GetStats() FontCacheStats

GetStats returns aggregated statistics across all shards

func (*OptimizedFontCache) Prefetch added in v1.0.1 

func (ofc *OptimizedFontCache) Prefetch(keys []string, compute func(key string) (*Font, error))

Prefetch warms up the cache with multiple keys concurrently

func (*OptimizedFontCache) Remove added in v1.0.1 

func (ofc *OptimizedFontCache) Remove(key string)

Remove removes a specific key from the cache

func (*OptimizedFontCache) Set added in v1.0.1 

func (ofc *OptimizedFontCache) Set(key string, font *Font)

Set stores a font in the cache

type OptimizedMemoryPool 

type OptimizedMemoryPool struct {
	// contains filtered or unexported fields
}

OptimizedMemoryPool provides better memory pool management

func NewOptimizedMemoryPool 

func NewOptimizedMemoryPool(size int) *OptimizedMemoryPool

NewOptimizedMemoryPool creates a pool with size tracking

func (*OptimizedMemoryPool) Get 

func (omp *OptimizedMemoryPool) Get() []byte

Get retrieves a buffer from the pool

func (*OptimizedMemoryPool) Put 

func (omp *OptimizedMemoryPool) Put(bufPtr *[]byte)

Put returns a buffer to the pool, resetting it

type OptimizedSorter 

type OptimizedSorter struct {
	// contains filtered or unexported fields
}

OptimizedSorter provides optimized sorting algorithms for large text collections

func NewOptimizedSorter 

func NewOptimizedSorter() *OptimizedSorter

NewOptimizedSorter creates a new optimized sorter

func (*OptimizedSorter) QuickSortTexts 

func (os *OptimizedSorter) QuickSortTexts(texts []Text, less func(i, j int) bool)

QuickSortTexts implements quicksort for text collections

func (*OptimizedSorter) SortTextHorizontalByOptimized 

func (os *OptimizedSorter) SortTextHorizontalByOptimized(th TextHorizontal)

SortTextHorizontalByOptimized sorts TextHorizontal using optimized algorithm

func (*OptimizedSorter) SortTextVerticalByOptimized 

func (os *OptimizedSorter) SortTextVerticalByOptimized(tv TextVertical)

SortTextVerticalByOptimized sorts TextVertical using optimized algorithm

func (*OptimizedSorter) SortTexts 

func (os *OptimizedSorter) SortTexts(texts []Text, less func(i, j int) bool)

SortTexts sorts a collection of texts using the most appropriate algorithm

func (*OptimizedSorter) SortTextsWithAlgorithm 

func (os *OptimizedSorter) SortTextsWithAlgorithm(texts []Text, less func(i, j int) bool, algorithm string)

SortTextsWithAlgorithm allows choosing a specific sorting algorithm

type OptimizedTextClusterSorter 

type OptimizedTextClusterSorter struct {
	// contains filtered or unexported fields
}

OptimizedTextClusterSorter provides optimized sorting for text clusters

func NewOptimizedTextClusterSorter 

func NewOptimizedTextClusterSorter() *OptimizedTextClusterSorter

NewOptimizedTextClusterSorter creates a new optimized cluster sorter

func (*OptimizedTextClusterSorter) SortTextBlocks 

func (otcs *OptimizedTextClusterSorter) SortTextBlocks(blocks []*TextBlock, sortBy string)

SortTextBlocks sorts text blocks by various criteria

type Outline 

type Outline struct {
	Title string    // title for this element
	Child []Outline // child elements
}

An Outline is a tree describing the outline (also known as the table of contents) of a document.

type PDFError 

type PDFError struct {
	Op   string // Operation that failed (e.g., "extract text", "parse font")
	Page int    // Page number where error occurred (0 if not page-specific)
	Path string // File path if applicable
	Err  error  // Underlying error
}

PDFError represents an error that occurred during PDF processing. It includes contextual information about where the error occurred.

func (*PDFError) Error 

func (e *PDFError) Error() string

func (*PDFError) Unwrap 

func (e *PDFError) Unwrap() error

type Page 

type Page struct {
	V Value
	// contains filtered or unexported fields
}

A Page represent a single page in a PDF file. The methods interpret a Page dictionary stored in V.

func (Page) ClassifyTextBlocks 

func (p Page) ClassifyTextBlocks() ([]ClassifiedBlock, error)

ClassifyTextBlocks is a convenience function that creates a classifier and runs classification

func (*Page) Cleanup added in v1.0.7 

func (p *Page) Cleanup()

Cleanup releases resources held by the Page, specifically the fontCache reference. Call this after processing a page to prevent memory leaks in batch operations. This method is safe to call multiple times.

func (Page) Content 

func (p Page) Content() Content

Content returns the page's content.

func (Page) Font 

func (p Page) Font(name string) Font

Font returns the font with the given name associated with the page.

func (Page) Fonts 

func (p Page) Fonts() []string

Fonts returns a list of the fonts associated with the page.

func (Page) GetPlainText 

func (p Page) GetPlainText(fonts map[string]*Font) (string, error)

GetPlainText returns the page's all text without format. fonts can be passed in (to improve parsing performance) or left nil

func (Page) GetPlainTextWithSmartOrdering 

func (p Page) GetPlainTextWithSmartOrdering(fonts map[string]*Font) (string, error)

GetPlainTextWithSmartOrdering extracts plain text using an improved text ordering algorithm that handles multi-column layouts and complex reading orders.

func (Page) GetTextByColumn 

func (p Page) GetTextByColumn() (Columns, error)

GetTextByColumn returns the page's all text grouped by column

func (Page) GetTextByRow 

func (p Page) GetTextByRow() (Rows, error)

GetTextByRow returns the page's all text grouped by rows

func (Page) OptimizedGetPlainText 

func (p Page) OptimizedGetPlainText(fonts map[string]*Font) (string, error)

OptimizedGetPlainText returns the page's all text using optimized string building. This version uses object pools and pre-allocation to reduce memory allocations.

func (Page) OptimizedGetTextByColumn 

func (p Page) OptimizedGetTextByColumn() (Columns, error)

OptimizedGetTextByColumn returns the page's all text grouped by column using optimized allocation

func (Page) OptimizedGetTextByRow 

func (p Page) OptimizedGetTextByRow() (Rows, error)

OptimizedGetTextByRow returns the page's all text grouped by rows using optimized allocation

func (Page) Resources 

func (p Page) Resources() Value

Resources returns the resources dictionary associated with the page.

func (*Page) SetFontCache added in v1.0.1 

func (p *Page) SetFontCache(cache *GlobalFontCache)

SetFontCache sets a font cache for this page to improve performance during text extraction by reusing parsed fonts. Deprecated: Use SetFontCacheInterface for better flexibility.

func (*Page) SetFontCacheInterface added in v1.0.1 

func (p *Page) SetFontCacheInterface(cache FontCacheInterface)

SetFontCacheInterface sets a font cache using the interface This supports both GlobalFontCache and OptimizedFontCache

type PageStream 

type PageStream struct {
	Page      Page
	PageNum   int
	HasText   bool
	TextCount int
}

PageStream represents a stream of pages

type ParallelExtractor added in v1.0.2 

type ParallelExtractor struct {
	// contains filtered or unexported fields
}

ParallelExtractor 并行提取器 结合所有优化的高级提取接口

Example (Basic)

ExampleParallelExtractor_basic 基本使用示例 

// 创建并行提取器
extractor := NewParallelExtractor(4) // 使用 4 个工作协程
defer extractor.Close()

// 注意：实际使用需要创建 Page 对象
// pages := []Page{...}

ctx := context.Background()

// 模拟空页面列表
var pages []Page

// 提取所有页面
results, err := extractor.ExtractAllPages(ctx, pages)
if err != nil {
	fmt.Printf("Error: %v\n", err)
	return
}

fmt.Printf("Extracted %d pages\n", len(results))

Output:

Extracted 0 pages

func NewParallelExtractor added in v1.0.2 

func NewParallelExtractor(workers int) *ParallelExtractor

NewParallelExtractor 创建并行提取器

func (*ParallelExtractor) Close added in v1.0.2 

func (pe *ParallelExtractor) Close()

Close 关闭并清理资源

func (*ParallelExtractor) ExtractAllPages added in v1.0.2 

func (pe *ParallelExtractor) ExtractAllPages(
	ctx context.Context,
	pages []Page,
) ([][]Text, error)

ExtractAllPages 提取所有页面（使用所有优化）

func (*ParallelExtractor) GetCacheStats added in v1.0.2 

func (pe *ParallelExtractor) GetCacheStats() ShardedCacheStats

GetCacheStats 获取缓存统计

func (*ParallelExtractor) GetPrefetchStats added in v1.0.2 

func (pe *ParallelExtractor) GetPrefetchStats() PrefetchStats

GetPrefetchStats 获取预取统计

type ParallelProcessor 

type ParallelProcessor struct {
	// contains filtered or unexported fields
}

ParallelProcessor handles multi-level parallel processing for PDF text extraction

func NewParallelProcessor 

func NewParallelProcessor(workers int) *ParallelProcessor

NewParallelProcessor creates a new parallel processor with the specified number of workers

func (*ParallelProcessor) ProcessPages 

func (pp *ParallelProcessor) ProcessPages(ctx context.Context, pages []Page, processorFunc func(Page) ([]Text, error)) ([][]Text, error)

ProcessPages processes multiple pages in parallel

func (*ParallelProcessor) ProcessTextBlocks 

func (pp *ParallelProcessor) ProcessTextBlocks(ctx context.Context, blocks []*TextBlock, processorFunc func(*TextBlock) (*TextBlock, error)) ([]*TextBlock, error)

ProcessTextBlocks processes multiple text blocks in parallel

func (*ParallelProcessor) ProcessTextInParallel 

func (pp *ParallelProcessor) ProcessTextInParallel(ctx context.Context, texts []Text, processorFunc func(Text) (Text, error)) ([]Text, error)

ProcessTextInParallel processes individual text elements in parallel

type ParallelTextExtractor 

type ParallelTextExtractor struct {
	// contains filtered or unexported fields
}

ParallelTextExtractor provides multi-level parallel extraction

func NewParallelTextExtractor 

func NewParallelTextExtractor(workers int) *ParallelTextExtractor

NewParallelTextExtractor creates a new parallel text extractor

func (*ParallelTextExtractor) ExtractWithParallelProcessing 

func (pte *ParallelTextExtractor) ExtractWithParallelProcessing(ctx context.Context, reader *Reader) ([]Text, error)

ExtractWithParallelProcessing extracts text using multi-level parallel processing

func (*ParallelTextExtractor) ParallelSort 

func (pte *ParallelTextExtractor) ParallelSort(ctx context.Context, texts []Text, less func(i, j int) bool) error

ParallelSort provides parallel sorting for large text collections

type PerformanceMetrics 

type PerformanceMetrics struct {
	ExtractDuration atomic.Int64 // 纳秒
	ParseDuration   atomic.Int64
	SortDuration    atomic.Int64
	TotalAllocs     atomic.Uint64
	BytesAllocated  atomic.Uint64
	GoroutineCount  atomic.Int32
	CacheHitRate    atomic.Uint64 // 百分比 * 100
}

PerformanceMetrics 性能指标收集器

func (*PerformanceMetrics) GetMetrics 

func (pm *PerformanceMetrics) GetMetrics() map[string]interface{}

GetMetrics 获取当前指标快照

func (*PerformanceMetrics) RecordAllocation 

func (pm *PerformanceMetrics) RecordAllocation(bytes uint64)

RecordAllocation 记录内存分配

func (*PerformanceMetrics) RecordExtractDuration 

func (pm *PerformanceMetrics) RecordExtractDuration(d time.Duration)

RecordExtractDuration 记录提取耗时

type Point 

type Point struct {
	X float64
	Y float64
}

A Point represents an X, Y pair.

type PoolStats added in v1.0.1 

type PoolStats struct {
	BucketSize int
	InUse      int // approximation, not perfectly accurate
}

GetStats returns statistics about pool usage (for debugging/monitoring)

type PoolWarmer added in v1.0.2 

type PoolWarmer struct {
	// contains filtered or unexported fields
}

PoolWarmer 内存池预热器 在应用启动时预先分配和填充内存池，减少运行时分配开销

func (*PoolWarmer) GetWarmupStats added in v1.0.2 

func (pw *PoolWarmer) GetWarmupStats() WarmupStats

GetWarmupStats 获取预热统计信息

func (*PoolWarmer) IsWarmed added in v1.0.2 

func (pw *PoolWarmer) IsWarmed() bool

IsWarmed 检查是否已预热

func (*PoolWarmer) Reset added in v1.0.2 

func (pw *PoolWarmer) Reset()

Reset 重置预热状态

func (*PoolWarmer) Warmup added in v1.0.2 

func (pw *PoolWarmer) Warmup(config *WarmupConfig) error

Warmup 执行内存池预热

type PrefetchItem added in v1.0.2 

type PrefetchItem struct {
	// contains filtered or unexported fields
}

PrefetchItem 预取项

type PrefetchQueue added in v1.0.2 

type PrefetchQueue struct {
	// contains filtered or unexported fields
}

PrefetchQueue 预取队列（优先级队列）

func (*PrefetchQueue) Len added in v1.0.2 

func (pq *PrefetchQueue) Len() int

func (*PrefetchQueue) Less added in v1.0.2 

func (pq *PrefetchQueue) Less(i, j int) bool

func (*PrefetchQueue) Pop added in v1.0.2 

func (pq *PrefetchQueue) Pop() interface{}

func (*PrefetchQueue) Push added in v1.0.2 

func (pq *PrefetchQueue) Push(x interface{})

func (*PrefetchQueue) Swap added in v1.0.2 

func (pq *PrefetchQueue) Swap(i, j int)

type PrefetchStats added in v1.0.2 

type PrefetchStats struct {
	PatternsTracked int
	QueueSize       int
	Enabled         bool
}

PrefetchStats 预取统计信息

type RTreeNode 

type RTreeNode struct {
	// contains filtered or unexported fields
}

RTreeNode represents a node in the R-tree

type RTreeSpatialIndex 

type RTreeSpatialIndex struct {
	// contains filtered or unexported fields
}

RTreeSpatialIndex provides a more sophisticated spatial index using a proper R-tree implementation

func NewRTreeSpatialIndex 

func NewRTreeSpatialIndex(texts []Text) *RTreeSpatialIndex

NewRTreeSpatialIndex creates a new R-tree based spatial index

func (*RTreeSpatialIndex) Insert 

func (rt *RTreeSpatialIndex) Insert(text Text)

Insert adds a text element to the R-tree

func (*RTreeSpatialIndex) Query 

func (rt *RTreeSpatialIndex) Query(bounds Rect) []Text

Query returns all text elements that intersect with the given bounds

type Reader 

type Reader struct {
	// contains filtered or unexported fields
}

A Reader is a single PDF file open for reading.

func NewReader 

func NewReader(f io.ReaderAt, size int64) (*Reader, error)

NewReader opens a file for reading, using the data in f with the given total size.

func NewReaderEncrypted 

func NewReaderEncrypted(f io.ReaderAt, size int64, pw func() string) (*Reader, error)

NewReaderEncrypted opens a file for reading, using the data in f with the given total size. If the PDF is encrypted, NewReaderEncrypted calls pw repeatedly to obtain passwords to try. If pw returns the empty string, NewReaderEncrypted stops trying to decrypt the file and returns an error.

func NewReaderEncryptedWithMmap 

func NewReaderEncryptedWithMmap(f io.ReaderAt, size int64, pw func() string) (*Reader, error)

NewReaderEncryptedWithMmap opens a file for reading with memory mapping for large files. If the file size exceeds 10MB, it uses memory mapping to reduce memory usage. This is a wrapper around NewReaderEncrypted that optimizes for large files.

func Open 

func Open(file string) (*os.File, *Reader, error)

Open opens a file for reading.

func (*Reader) BatchExtractText 

func (r *Reader) BatchExtractText(pageNums []int, useLazy bool) (map[int]string, error)

BatchExtractText extracts text from multiple pages using lazy loading and object pooling This is optimized for processing many pages without keeping all in memory

func (*Reader) ClearCache added in v1.0.6 

func (r *Reader) ClearCache()

ClearCache clears the object cache, releasing all cached objects. This is useful for freeing memory after batch processing large PDFs.

func (*Reader) Close added in v1.0.2 

func (r *Reader) Close() error

Close closes the Reader and releases associated resources. If the underlying ReaderAt implements io.Closer, it will be closed.

func (*Reader) ExtractAllPagesParallel added in v1.0.2 

func (r *Reader) ExtractAllPagesParallel(ctx context.Context, workers int) ([]string, error)

ExtractAllPagesParallel 使用增强的并行提取器提取所有页面文本 这个方法集成了所有性能优化：分片缓存、字体预取、零拷贝等

Example

ExampleReader_ExtractAllPagesParallel 使用 Reader 的并行提取方法 

// 注意：这个示例需要实际的 PDF 文件
// 这里仅展示 API 用法

/*
	// 打开 PDF 文件
	f, r, err := Open("document.pdf")
	if err != nil {
		panic(err)
	}
	defer f.Close()

	// 创建上下文
	ctx, cancel := context.WithTimeout(context.Background(), 1*time.Minute)
	defer cancel()

	// 并行提取所有页面文本
	pages, err := r.ExtractAllPagesParallel(ctx, 0) // 0 = 自动检测 CPU 核心数
	if err != nil {
		panic(err)
	}

	// 输出每页的文本
	for i, pageText := range pages {
		fmt.Printf("Page %d has %d characters\n", i+1, len(pageText))
	}
*/

func (*Reader) ExtractPagesBatch added in v1.0.1 

func (r *Reader) ExtractPagesBatch(opts BatchExtractOptions) <-chan BatchResult

ExtractPagesBatch extracts text from multiple pages in batches This is optimized for high-throughput scenarios with many pages

Example 
// This example shows how to use batch extraction
// (requires a real PDF file to run)

// r, err := Open("document.pdf")
// if err != nil {
//     log.Fatal(err)
// }
// defer r.Close()
//
// opts := BatchExtractOptions{
//     Workers: 4,
//     Pages:   []int{1, 2, 3, 4, 5}, // Extract first 5 pages
// }
//
// for result := range r.ExtractPagesBatch(opts) {
//     if result.Error != nil {
//         log.Printf("Error on page %d: %v", result.PageNum, result.Error)
//         continue
//     }
//     fmt.Printf("Page %d: %d characters\n", result.PageNum, len(result.Text))
// }

func (*Reader) ExtractPagesBatchToString added in v1.0.1 

func (r *Reader) ExtractPagesBatchToString(opts BatchExtractOptions) (string, error)

ExtractPagesBatchToString is a convenience function that collects all results into a single string

Example 
// This example shows how to extract all pages to a single string

// r, err := Open("document.pdf")
// if err != nil {
//     log.Fatal(err)
// }
// defer r.Close()
//
// opts := BatchExtractOptions{
//     Workers:       8,
//     SmartOrdering: true,
// }
//
// text, err := r.ExtractPagesBatchToString(opts)
// if err != nil {
//     log.Fatal(err)
// }
//
// fmt.Printf("Extracted %d characters from %d pages\n", len(text), r.NumPage())

func (*Reader) ExtractStructuredBatch added in v1.0.1 

func (r *Reader) ExtractStructuredBatch(opts BatchExtractOptions) <-chan StructuredBatchResult

ExtractStructuredBatch extracts structured text in batches

func (*Reader) ExtractWithContext 

func (r *Reader) ExtractWithContext(ctx context.Context, opts ExtractOptions) (io.Reader, error)

ExtractWithContext extracts plain text from all pages with cancellation support

func (*Reader) GetCacheCapacity added in v1.0.6 

func (r *Reader) GetCacheCapacity() int

GetCacheCapacity returns the current object cache capacity. Returns 0 if no capacity limit is set (unbounded cache).

func (*Reader) GetMetadata 

func (r *Reader) GetMetadata() (Metadata, error)

GetMetadata extracts metadata from the PDF document

func (*Reader) GetPlainText 

func (r *Reader) GetPlainText() (reader io.Reader, err error)

GetPlainText returns all the text in the PDF file

func (*Reader) GetPlainTextConcurrent 

func (r *Reader) GetPlainTextConcurrent(workers int) (io.Reader, error)

GetPlainTextConcurrent extracts all pages concurrently using the specified number of workers.

func (*Reader) GetStyledTexts 

func (r *Reader) GetStyledTexts() (sentences []Text, err error)

GetStyledTexts returns list all sentences in an array, that are included styles

func (*Reader) NumPage 

func (r *Reader) NumPage() int

NumPage returns the number of pages in the PDF file.

func (*Reader) Outline 

func (r *Reader) Outline() Outline

Outline returns the document outline. The Outline returned is the root of the outline tree and typically has no Title itself. That is, the children of the returned root are the top-level entries in the outline.

func (*Reader) Page 

func (r *Reader) Page(num int) Page

Page returns the page for the given page number. Page numbers are indexed starting at 1, not 0. If the page is not found, Page returns a Page with p.V.IsNull().

func (*Reader) SetCacheCapacity 

func (r *Reader) SetCacheCapacity(n int)

func (*Reader) SetMetadata 

func (r *Reader) SetMetadata(meta Metadata) error

SetMetadata sets metadata fields in the PDF (for future write support) Currently not implemented as the library is read-only

func (*Reader) Trailer 

func (r *Reader) Trailer() Value

Trailer returns the file's Trailer value.

type Rect 

type Rect struct {
	Min, Max Point
}

A Rect represents a rectangle.

type ResourceManager 

type ResourceManager struct {
	// contains filtered or unexported fields
}

ResourceManager provides automatic resource cleanup

func NewResourceManager 

func NewResourceManager() *ResourceManager

NewResourceManager creates a new resource manager

func (*ResourceManager) Add 

func (rm *ResourceManager) Add(resource io.Closer)

Add adds a resource to be managed

func (*ResourceManager) Close 

func (rm *ResourceManager) Close() error

Close closes all managed resources

type ResultCache 

type ResultCache struct {
	// contains filtered or unexported fields
}

ResultCache provides caching for parsed and classified results

func GetGlobalCache 

func GetGlobalCache() *ResultCache

GetGlobalCache returns a singleton cache instance

func NewResultCache 

func NewResultCache(maxSize int64, ttl time.Duration, policy string) *ResultCache

NewResultCache creates a new result cache with specified parameters

func (*ResultCache) Clear 

func (rc *ResultCache) Clear()

Clear removes all items from the cache

func (*ResultCache) Close added in v1.0.5 

func (rc *ResultCache) Close()

Close stops the cleanup goroutine and releases resources

func (*ResultCache) Get 

func (rc *ResultCache) Get(key string) (interface{}, bool)

Get retrieves an item from the cache

func (*ResultCache) GetHitRatio 

func (rc *ResultCache) GetHitRatio() float64

GetHitRatio returns the cache hit ratio

func (*ResultCache) GetStats 

func (rc *ResultCache) GetStats() CacheStats

GetStats returns cache statistics

func (*ResultCache) Has 

func (rc *ResultCache) Has(key string) bool

Has checks if a key exists in the cache (without updating access stats)

func (*ResultCache) Put 

func (rc *ResultCache) Put(key string, value interface{})

Put adds an item to the cache

func (*ResultCache) Remove 

func (rc *ResultCache) Remove(key string) bool

Remove removes an item from the cache

type Row 

type Row struct {
	Position int64
	Content  TextHorizontal
}

Row represents the contents of a row

type Rows 

type Rows []*Row

Rows is a list of rows

type ShardedCache added in v1.0.2 

type ShardedCache struct {
	// contains filtered or unexported fields
}

ShardedCache 实现了一个高性能分片缓存，具有以下特性： - 256 个分片以最小化锁竞争 - 每个分片独立的锁和 LRU 链表 - 原子操作实现的统计信息 - 自适应淘汰策略

func NewShardedCache added in v1.0.2 

func NewShardedCache(maxSize int, ttl time.Duration) *ShardedCache

NewShardedCache 创建新的分片缓存

func (*ShardedCache) Clear added in v1.0.2 

func (sc *ShardedCache) Clear()

Clear 清空所有缓存

func (*ShardedCache) Close added in v1.0.5 

func (sc *ShardedCache) Close()

Close 停止清理 goroutine 并释放资源

func (*ShardedCache) Delete added in v1.0.2 

func (sc *ShardedCache) Delete(key string)

Delete 删除缓存条目

func (*ShardedCache) Get added in v1.0.2 

func (sc *ShardedCache) Get(key string) (interface{}, bool)

Get 从缓存获取值

func (*ShardedCache) GetStats added in v1.0.2 

func (sc *ShardedCache) GetStats() ShardedCacheStats

GetStats 获取缓存统计信息

func (*ShardedCache) Set added in v1.0.2 

func (sc *ShardedCache) Set(key string, value interface{}, size int64)

Set 设置缓存值

type ShardedCacheEntry added in v1.0.2 

type ShardedCacheEntry struct {
	// contains filtered or unexported fields
}

ShardedCacheEntry 代表缓存条目

type ShardedCacheStats added in v1.0.2 

type ShardedCacheStats struct {
	Hits      uint64
	Misses    uint64
	Evictions uint64
	Entries   int64
	Size      int64
}

ShardedCacheStats 缓存统计信息

type SizedBytePool added in v1.0.1 

type SizedBytePool struct {
	// contains filtered or unexported fields
}

SizedBytePool implements a multi-level size-bucketed object pool for byte slices. It reduces memory allocation overhead by reusing buffers of appropriate sizes.

Size buckets: 16B, 32B, 64B, 128B, 256B, 512B, 1KB, 4KB

func NewSizedBytePool added in v1.0.1 

func NewSizedBytePool() *SizedBytePool

NewSizedBytePool creates a new sized byte pool with 8 size buckets

func (*SizedBytePool) Get added in v1.0.1 

func (sp *SizedBytePool) Get(size int) []byte

Get retrieves a byte slice from the appropriate size bucket Returns a buffer with at least the requested capacity

func (*SizedBytePool) Put added in v1.0.1 

func (sp *SizedBytePool) Put(buf []byte)

Put returns a byte slice to the appropriate pool The slice is cleared before being returned to the pool

type SizedPool 

type SizedPool struct {
	// contains filtered or unexported fields
}

1. 细粒度对象池 - 多级大小分桶

func NewSizedPool 

func NewSizedPool() *SizedPool

func (*SizedPool) Get 

func (sp *SizedPool) Get(size int) []byte

func (*SizedPool) Put 

func (sp *SizedPool) Put(bufPtr *[]byte)

type SizedTextSlicePool added in v1.0.1 

type SizedTextSlicePool struct {
	// contains filtered or unexported fields
}

SizedTextSlicePool implements a size-bucketed pool for Text slices Similar to SizedBytePool but for []Text instead of []byte

func NewSizedTextSlicePool added in v1.0.1 

func NewSizedTextSlicePool() *SizedTextSlicePool

NewSizedTextSlicePool creates a new sized text slice pool Buckets: 8, 16, 32, 64, 128, 256 texts

func (*SizedTextSlicePool) Get added in v1.0.1 

func (sp *SizedTextSlicePool) Get(size int) []Text

Get retrieves a Text slice from the appropriate size bucket

func (*SizedTextSlicePool) Put added in v1.0.1 

func (sp *SizedTextSlicePool) Put(slice []Text)

Put returns a Text slice to the appropriate pool

type SortStrategy added in v1.0.1 

type SortStrategy int

SortStrategy represents different sorting algorithms available 

const (
	StrategyAuto      SortStrategy = iota // Automatically select best algorithm
	StrategyRadix                         // Radix sort for numeric keys
	StrategyQuick                         // Quicksort for general comparison
	StrategyInsertion                     // Insertion sort for small arrays
	StrategyStandard                      // Go standard library sort
)

type SortingMetrics added in v1.0.1 

type SortingMetrics struct {
	RadixSortCount     int
	QuickSortCount     int
	InsertionSortCount int
	StandardSortCount  int
}

SortingMetrics tracks performance of different sorting strategies

func GetSortingMetrics added in v1.0.1 

func GetSortingMetrics() SortingMetrics

GetSortingMetrics returns current sorting metrics

type SpatialIndex 

type SpatialIndex struct {
	// contains filtered or unexported fields
}

SpatialIndex provides spatial indexing for efficient text location queries This is a simple implementation using a grid-based approach; for production use, consider a more sophisticated structure like R-tree

func NewSpatialIndex 

func NewSpatialIndex(texts []Text) *SpatialIndex

NewSpatialIndex creates a new spatial index from text elements

func (*SpatialIndex) Query 

func (si *SpatialIndex) Query(bounds Rect) []Text

Query returns all text elements that potentially intersect with the given bounds

type SpatialIndexInterface 

type SpatialIndexInterface interface {
	Query(bounds Rect) []Text
	Insert(text Text)
}

SpatialIndex interface to allow using either grid or R-tree implementation

func NewSpatialIndexInterface 

func NewSpatialIndexInterface(texts []Text) SpatialIndexInterface

NewSpatialIndexInterface creates a spatial index interface (can be switched between implementations)

type Stack 

type Stack struct {
	// contains filtered or unexported fields
}

A Stack represents a stack of values.

func (*Stack) Len 

func (stk *Stack) Len() int

func (*Stack) Pop 

func (stk *Stack) Pop() Value

func (*Stack) Push 

func (stk *Stack) Push(v Value)

type StartupConfig added in v1.0.2 

type StartupConfig struct {
	WarmupPools       bool
	WarmupConfig      *WarmupConfig
	PreallocateCaches bool
	FontCacheSize     int
	ResultCacheSize   int
	TuneGC            bool
	GCPercent         int
	MemoryBallast     int64
	SetMaxProcs       bool
	MaxProcs          int
}

StartupConfig 启动配置

func DefaultStartupConfig added in v1.0.2 

func DefaultStartupConfig() *StartupConfig

DefaultStartupConfig 默认启动配置

type StreamProcessor 

type StreamProcessor struct {
	// contains filtered or unexported fields
}

StreamProcessor handles streaming processing of PDF content to minimize memory usage

func NewStreamProcessor 

func NewStreamProcessor(chunkSize, bufferSize int, maxMemory int64) *StreamProcessor

NewStreamProcessor creates a new streaming processor

func (*StreamProcessor) Close 

func (sp *StreamProcessor) Close()

Close releases resources used by the stream processor

func (*StreamProcessor) ProcessPageStream 

func (sp *StreamProcessor) ProcessPageStream(reader *Reader, handler func(PageStream) error) error

ProcessPageStream processes pages in a streaming fashion

func (*StreamProcessor) ProcessTextBlockStream 

func (sp *StreamProcessor) ProcessTextBlockStream(reader *Reader, handler func(TextBlockStream) error) error

ProcessTextBlockStream processes text blocks in a streaming fashion

func (*StreamProcessor) ProcessTextStream 

func (sp *StreamProcessor) ProcessTextStream(reader *Reader, handler func(TextStream) error) error

ProcessTextStream processes text in a streaming fashion

type StreamingBatchExtractor added in v1.0.1 

type StreamingBatchExtractor struct {
	// contains filtered or unexported fields
}

StreamingBatchExtractor provides a streaming interface for batch extraction This is useful for very large PDFs where you want to process results as they arrive

Example 
// This example shows streaming batch extraction with a callback

// r, err := Open("document.pdf")
// if err != nil {
//     log.Fatal(err)
// }
// defer r.Close()
//
// ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
// defer cancel()
//
// opts := BatchExtractOptions{
//     Context: ctx,
//     Workers: 4,
// }
//
// extractor := NewStreamingBatchExtractor(r, opts)
// extractor.Start()
//
// err = extractor.ProcessAll(func(result BatchResult) error {
//     if result.Error != nil {
//         return result.Error
//     }
//     // Process each page as it arrives
//     fmt.Printf("Processing page %d...\n", result.PageNum)
//     return nil
// })
//
// if err != nil {
//     log.Fatal(err)
// }

func NewStreamingBatchExtractor added in v1.0.1 

func NewStreamingBatchExtractor(r *Reader, opts BatchExtractOptions) *StreamingBatchExtractor

NewStreamingBatchExtractor creates a new streaming batch extractor

func (*StreamingBatchExtractor) Next added in v1.0.1 

func (sbe *StreamingBatchExtractor) Next() *BatchResult

Next returns the next result, or nil if done

func (*StreamingBatchExtractor) ProcessAll added in v1.0.1 

func (sbe *StreamingBatchExtractor) ProcessAll(callback func(BatchResult) error) error

ProcessAll processes all pages with a callback function

func (*StreamingBatchExtractor) Start added in v1.0.1 

func (sbe *StreamingBatchExtractor) Start()

Start begins the extraction process

type StreamingMetadataExtractor 

type StreamingMetadataExtractor struct {
	// contains filtered or unexported fields
}

StreamingMetadataExtractor extracts metadata in a streaming fashion

func NewStreamingMetadataExtractor 

func NewStreamingMetadataExtractor(chunkSize, bufferSize int, maxMemory int64) *StreamingMetadataExtractor

NewStreamingMetadataExtractor creates a new streaming metadata extractor

func (*StreamingMetadataExtractor) ExtractMetadataStream 

func (sme *StreamingMetadataExtractor) ExtractMetadataStream(reader *Reader) (<-chan Metadata, <-chan error)

ExtractMetadataStream extracts metadata in a streaming way

type StreamingTextClassifier 

type StreamingTextClassifier struct {
	// contains filtered or unexported fields
}

StreamingTextClassifier classifies text in a streaming fashion to minimize memory usage

func NewStreamingTextClassifier 

func NewStreamingTextClassifier(chunkSize, bufferSize int, maxMemory int64) *StreamingTextClassifier

NewStreamingTextClassifier creates a new streaming text classifier

func (*StreamingTextClassifier) ClassifyTextStream 

func (stc *StreamingTextClassifier) ClassifyTextStream(reader *Reader) (<-chan ClassifiedBlock, <-chan error)

ClassifyTextStream classifies text in a streaming way

type StreamingTextExtractor 

type StreamingTextExtractor struct {
	// contains filtered or unexported fields
}

StreamingTextExtractor provides memory-efficient text extraction for large PDFs

func NewStreamingTextExtractor 

func NewStreamingTextExtractor(r *Reader, maxCachedPages int) *StreamingTextExtractor

NewStreamingTextExtractor creates a streaming extractor for large PDFs

func (*StreamingTextExtractor) Close 

func (e *StreamingTextExtractor) Close()

Close releases resources used by the extractor

func (*StreamingTextExtractor) GetProgress 

func (e *StreamingTextExtractor) GetProgress() float64

GetProgress returns the extraction progress (0.0 to 1.0)

func (*StreamingTextExtractor) NextBatch 

func (e *StreamingTextExtractor) NextBatch() (results map[int]string, hasMore bool, err error)

NextBatch extracts text from the next batch of pages

func (*StreamingTextExtractor) NextPage 

func (e *StreamingTextExtractor) NextPage() (pageNum int, text string, hasMore bool, err error)

NextPage extracts text from the next page

func (*StreamingTextExtractor) Reset 

func (e *StreamingTextExtractor) Reset()

Reset resets the extractor to the beginning

type StringBuffer added in v1.0.2 

type StringBuffer struct {
	// contains filtered or unexported fields
}

StringBuffer 字符串构建缓冲区，优化多次拼接

Example

ExampleStringBuffer 演示 StringBuffer 的使用 

builder := NewStringBuffer(100)

builder.WriteString("Hello")
builder.WriteByte(' ')
builder.WriteString("World")

result := builder.StringCopy()
fmt.Println(result)

Output:

Hello World

func NewStringBuffer added in v1.0.2 

func NewStringBuffer(capacity int) *StringBuffer

NewStringBuffer 创建新的字符串缓冲区

func (*StringBuffer) Bytes added in v1.0.2 

func (sb *StringBuffer) Bytes() []byte

Bytes 返回底层字节切片

func (*StringBuffer) Cap added in v1.0.2 

func (sb *StringBuffer) Cap() int

Cap 返回容量

func (*StringBuffer) Len added in v1.0.2 

func (sb *StringBuffer) Len() int

Len 返回当前长度

func (*StringBuffer) Reset added in v1.0.2 

func (sb *StringBuffer) Reset()

Reset 重置缓冲区

func (*StringBuffer) String added in v1.0.2 

func (sb *StringBuffer) String() string

String 零拷贝返回字符串 警告：返回后不要再使用 StringBuffer

func (*StringBuffer) StringCopy added in v1.0.2 

func (sb *StringBuffer) StringCopy() string

StringCopy 安全返回字符串副本

func (*StringBuffer) WriteByte added in v1.0.2 

func (sb *StringBuffer) WriteByte(b byte) error

WriteByte 写入单个字节

func (*StringBuffer) WriteBytes added in v1.0.2 

func (sb *StringBuffer) WriteBytes(b []byte)

WriteBytes 写入字节切片

func (*StringBuffer) WriteString added in v1.0.2 

func (sb *StringBuffer) WriteString(s string)

WriteString 写入字符串

type StringBuilderPool added in v1.0.1 

type StringBuilderPool struct {
	// contains filtered or unexported fields
}

StringBuilderPool provides size-aware string builder pooling

type StringPool added in v1.0.2 

type StringPool struct {
	// contains filtered or unexported fields
}

StringPool 字符串池，复用常见字符串

Example

ExampleStringPool 演示字符串池的使用 

pool := NewStringPool()

// 常用字符串放入池中
fontName1 := pool.Intern("Arial")
fontName2 := pool.Intern("Arial") // 重复的字符串会复用

fmt.Println(fontName1 == fontName2) // 指针相等
fmt.Println(pool.Size())

Output:

true
1

func NewStringPool added in v1.0.2 

func NewStringPool() *StringPool

NewStringPool 创建新的字符串池

func (*StringPool) Clear added in v1.0.2 

func (sp *StringPool) Clear()

Clear 清空池

func (*StringPool) Intern added in v1.0.2 

func (sp *StringPool) Intern(s string) string

Intern 将字符串加入池中并返回池化版本 相同内容的字符串将共享内存

func (*StringPool) Size added in v1.0.2 

func (sp *StringPool) Size() int

Size 返回池中字符串数量

type StructuredBatchResult added in v1.0.1 

type StructuredBatchResult struct {
	PageNum int
	Blocks  []ClassifiedBlock
	Error   error
}

BatchExtractStructured extracts structured text from multiple pages in batches

type Task 

type Task interface {
	Execute() error
}

Task 任务接口

type Text 

type Text struct {
	Font      string  // the font used
	FontSize  float64 // the font size, in points (1/72 of an inch)
	X         float64 // the X coordinate, in points, increasing left to right
	Y         float64 // the Y coordinate, in points, increasing bottom to top
	W         float64 // the width of the text, in points
	S         string  // the actual UTF-8 text
	Vertical  bool    // whether the text is drawn vertically
	Bold      bool    // whether the text is bold
	Italic    bool    // whether the text is italic
	Underline bool    // whether the text is underlined
}

A Text represents a single piece of text drawn on a page.

func GetSizedTextSlice added in v1.0.1 

func GetSizedTextSlice(size int) []Text

GetSizedTextSlice retrieves a Text slice from the global pool

func GetText 

func GetText() *Text

GetText retrieves a Text object from the appropriate pool based on content size

func GetTextBySize 

func GetTextBySize(contentLength int) *Text

GetTextBySize retrieves a Text object from the appropriate pool based on content size

func GetTextSlice 

func GetTextSlice() []Text

GetTextSlice retrieves a Text slice from the pool

type TextBlock 

type TextBlock struct {
	Texts       []Text
	MinX        float64
	MaxX        float64
	MinY        float64
	MaxY        float64
	AvgFontSize float64
}

TextBlock represents a coherent block of text (like a paragraph or column)

func ClusterTextBlocksOptimized 

func ClusterTextBlocksOptimized(texts []Text) []*TextBlock

ClusterTextBlocksOptimized 使用KD树优化的文本块聚类 优化版本:减少临时对象分配,使用对象池

func (*TextBlock) Bounds 

func (tb *TextBlock) Bounds() Rect

Bounds returns the bounding box of the text block

func (*TextBlock) Center 

func (tb *TextBlock) Center() Point

Center returns the center point of the text block

func (*TextBlock) Height 

func (tb *TextBlock) Height() float64

Height returns the height of the text block

func (*TextBlock) Width 

func (tb *TextBlock) Width() float64

Width returns the width of the text block

type TextBlockStream 

type TextBlockStream struct {
	Block   *TextBlock
	PageNum int
	Type    BlockType
	Level   int
	Text    string
}

TextBlockStream represents a stream of text blocks

type TextClassifier 

type TextClassifier struct {
	// contains filtered or unexported fields
}

TextClassifier classifies text runs into semantic blocks

func NewTextClassifier 

func NewTextClassifier(texts []Text, pageWidth, pageHeight float64) *TextClassifier

NewTextClassifier creates a new text classifier

func (*TextClassifier) ClassifyBlocks 

func (tc *TextClassifier) ClassifyBlocks() []ClassifiedBlock

ClassifyBlocks classifies text runs into semantic blocks

type TextEncoding 

type TextEncoding interface {
	// Decode returns the UTF-8 text corresponding to
	// the sequence of code points in raw.
	Decode(raw string) (text string)
}

A TextEncoding represents a mapping between font code points and UTF-8 text.

type TextHorizontal 

type TextHorizontal []Text

TextHorizontal implements sort.Interface for sorting a slice of Text values in horizontal order, left to right, and then top to bottom within a column.

func (TextHorizontal) Len 

func (x TextHorizontal) Len() int

func (TextHorizontal) Less 

func (x TextHorizontal) Less(i, j int) bool

func (TextHorizontal) Swap 

func (x TextHorizontal) Swap(i, j int)

type TextStream 

type TextStream struct {
	Text       string
	PageNum    int
	Font       string
	FontSize   float64
	X, Y       float64
	W          float64
	Vertical   bool
	Confidence float64 // Confidence in the text recognition (0-1)
}

TextStream represents a stream of text with metadata

type TextVertical 

type TextVertical []Text

TextVertical implements sort.Interface for sorting a slice of Text values in vertical order, top to bottom, and then left to right within a line.

func (TextVertical) Len 

func (x TextVertical) Len() int

func (TextVertical) Less 

func (x TextVertical) Less(i, j int) bool

func (TextVertical) Swap 

func (x TextVertical) Swap(i, j int)

type TextWithLanguage 

type TextWithLanguage struct {
	Text       Text
	Language   LanguageInfo
	Confidence float64
}

TextWithLanguage represents text with detected language information

type Value 

type Value struct {
	// contains filtered or unexported fields
}

A Value is a single PDF value, such as an integer, dictionary, or array. The zero Value is a PDF null (Kind() == Null, IsNull() = true).

func (Value) Bool 

func (v Value) Bool() bool

Bool returns v's boolean value. If v.Kind() != Bool, Bool returns false.

func (Value) Float64 

func (v Value) Float64() float64

Float64 returns v's float64 value, converting from integer if necessary. If v.Kind() != Float64 and v.Kind() != Int64, Float64 returns 0.

func (Value) Index 

func (v Value) Index(i int) Value

Index returns the i'th element in the array v. If v.Kind() != Array or if i is outside the array bounds, Index returns a null Value.

func (Value) Int64 

func (v Value) Int64() int64

Int64 returns v's int64 value. If v.Kind() != Int64, Int64 returns 0.

func (Value) IsNull 

func (v Value) IsNull() bool

IsNull reports whether the value is a null. It is equivalent to Kind() == Null.

func (Value) Key 

func (v Value) Key(key string) Value

Key returns the value associated with the given name key in the dictionary v. Like the result of the Name method, the key should not include a leading slash. If v is a stream, Key applies to the stream's header dictionary. If v.Kind() != Dict and v.Kind() != Stream, Key returns a null Value.

func (Value) Keys 

func (v Value) Keys() []string

Keys returns a sorted list of the keys in the dictionary v. If v is a stream, Keys applies to the stream's header dictionary. If v.Kind() != Dict and v.Kind() != Stream, Keys returns nil.

func (Value) Kind 

func (v Value) Kind() ValueKind

Kind reports the kind of value underlying v.

func (Value) Len 

func (v Value) Len() int

Len returns the length of the array v. If v.Kind() != Array, Len returns 0.

func (Value) Name 

func (v Value) Name() string

Name returns v's name value. If v.Kind() != Name, Name returns the empty string. The returned name does not include the leading slash: if v corresponds to the name written using the syntax /Helvetica, Name() == "Helvetica".

func (Value) RawString 

func (v Value) RawString() string

RawString returns v's string value. If v.Kind() != String, RawString returns the empty string.

func (Value) Reader 

func (v Value) Reader() io.ReadCloser

Reader returns the data contained in the stream v. If v.Kind() != Stream, Reader returns a ReadCloser that responds to all reads with a “stream not present” error.

func (Value) String 

func (v Value) String() string

String returns a textual representation of the value v. Note that String is not the accessor for values with Kind() == String. To access such values, see RawString, Text, and TextFromUTF16.

func (Value) Text 

func (v Value) Text() string

Text returns v's string value interpreted as a “text string” (defined in the PDF spec) and converted to UTF-8. If v.Kind() != String, Text returns the empty string.

func (Value) TextFromUTF16 

func (v Value) TextFromUTF16() string

TextFromUTF16 returns v's string value interpreted as big-endian UTF-16 and then converted to UTF-8. If v.Kind() != String or if the data is not valid UTF-16, TextFromUTF16 returns the empty string.

type ValueKind 

type ValueKind int

A ValueKind specifies the kind of data underlying a Value. 

const (
	Null ValueKind = iota
	Bool
	Integer
	Real
	String
	Name
	Dict
	Array
	Stream
)

The PDF value kinds.

type WSDeque 

type WSDeque struct {
	// contains filtered or unexported fields
}

5. Work-Stealing Deque (Chase-Lev算法)

func NewWSDeque 

func NewWSDeque(size int) *WSDeque

func (*WSDeque) PopBottom 

func (d *WSDeque) PopBottom() WSTask

PopBottom - owner线程从底部弹出（LIFO）

func (*WSDeque) PushBottom 

func (d *WSDeque) PushBottom(task WSTask)

PushBottom - owner线程从底部插入

func (*WSDeque) Steal 

func (d *WSDeque) Steal() WSTask

Steal - 其他线程从顶部偷取（FIFO）

type WSTask 

type WSTask interface {
	Execute()
}

type WSWorker 

type WSWorker struct {
	// contains filtered or unexported fields
}

type WarmupConfig added in v1.0.2 

type WarmupConfig struct {
	// BytePoolWarmup 每个大小桶预热的缓冲区数量
	BytePoolWarmup map[int]int

	// TextPoolWarmup 每个大小桶预热的文本切片数量
	TextPoolWarmup map[int]int

	// Concurrent 是否并发预热
	Concurrent bool

	// MaxGoroutines 最大并发 goroutine 数
	MaxGoroutines int
}

WarmupConfig 预热配置

func AggressiveWarmupConfig added in v1.0.2 

func AggressiveWarmupConfig() *WarmupConfig

AggressiveWarmupConfig 返回激进的预热配置（更多预分配）

func DefaultWarmupConfig added in v1.0.2 

func DefaultWarmupConfig() *WarmupConfig

DefaultWarmupConfig 返回默认预热配置

func LightWarmupConfig added in v1.0.2 

func LightWarmupConfig() *WarmupConfig

LightWarmupConfig 返回轻量预热配置（较少预分配）

type WarmupStats added in v1.0.2 

type WarmupStats struct {
	BytePoolSizes  map[int]int
	TextPoolSizes  map[int]int
	TotalAllocated int64
	IsWarmed       bool
}

WarmupStats 预热统计信息

type WorkStealingExecutor 

type WorkStealingExecutor struct {
	// contains filtered or unexported fields
}

6. Work-Stealing线程池

func NewWorkStealingExecutor 

func NewWorkStealingExecutor(numWorkers int) *WorkStealingExecutor

func (*WorkStealingExecutor) Start 

func (p *WorkStealingExecutor) Start()

func (*WorkStealingExecutor) Stop 

func (p *WorkStealingExecutor) Stop()

func (*WorkStealingExecutor) Submit 

func (p *WorkStealingExecutor) Submit(task WSTask)

type WorkStealingScheduler 

type WorkStealingScheduler struct {
	// contains filtered or unexported fields
}

WorkStealingScheduler 工作窃取调度器 减少goroutine创建开销，提升并行处理效率

func NewWorkStealingScheduler 

func NewWorkStealingScheduler(numWorkers int) *WorkStealingScheduler

NewWorkStealingScheduler 创建工作窃取调度器

func (*WorkStealingScheduler) Start 

func (wss *WorkStealingScheduler) Start()

Start 启动调度器

func (*WorkStealingScheduler) Stop 

func (wss *WorkStealingScheduler) Stop()

Stop 停止调度器

func (*WorkStealingScheduler) Submit 

func (wss *WorkStealingScheduler) Submit(task Task)

Submit 提交任务

func (*WorkStealingScheduler) Wait 

func (wss *WorkStealingScheduler) Wait()

Wait 等待所有任务完成

type Worker 

type Worker struct {
	// contains filtered or unexported fields
}

Worker 工作线程

type WorkerPool added in v1.0.2 

type WorkerPool struct {
	// contains filtered or unexported fields
}

WorkerPool 工作池

func (*WorkerPool) GetStats added in v1.0.2 

func (wp *WorkerPool) GetStats() WorkerPoolStats

GetStats 获取工作池统计信息

type WorkerPoolStats added in v1.0.2 

type WorkerPoolStats struct {
	Workers    int
	ActiveJobs int64
	TotalJobs  int64
}

WorkerPoolStats 工作池统计信息

type YBand 

type YBand struct {
	MinY, MaxY float64
	Blocks     []*TextBlock
}

YBand represents a horizontal band of text on a page

type ZeroCopyBuilder 

type ZeroCopyBuilder struct {
	// contains filtered or unexported fields
}

2. 零拷贝字符串构建器

func NewZeroCopyBuilder 

func NewZeroCopyBuilder(cap int) *ZeroCopyBuilder

func (*ZeroCopyBuilder) Reset 

func (b *ZeroCopyBuilder) Reset()

func (*ZeroCopyBuilder) UnsafeString 

func (b *ZeroCopyBuilder) UnsafeString() string

UnsafeString 零拷贝返回字符串（注意：底层buffer不能修改）

func (*ZeroCopyBuilder) WriteByte 

func (b *ZeroCopyBuilder) WriteByte(c byte) error

func (*ZeroCopyBuilder) WriteString 

func (b *ZeroCopyBuilder) WriteString(s string)

Notes

Bugs

  • The package is incomplete, although it has been used successfully on some large real-world PDF files.

  • The library makes no attempt at efficiency beyond the value cache and font cache. Further optimizations could improve performance for large files.

  • The support for reading encrypted files is limited to basic RC4 and AES encryption.

Source Files

View all Source files

Directories

Path Synopsis
cmd
pdfcli command
test_coords command
test_ordering command
examples
batch_fontcache command
extract command
Example: Extract text from a PDF file with various methods
 Example: Extract text from a PDF file with various methods
extract_text_performance command
performance command
Example demonstrating performance optimization features
 Example demonstrating performance optimization features
smart_ordering command
Pdfpasswd searches for the password for an encrypted PDF by trying all strings over a given alphabet up to a given length.
 Pdfpasswd searches for the password for an encrypted PDF by trying all strings over a given alphabet up to a given length.

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL
go.dev uses cookies from Google to deliver and enhance the quality of its services and to analyze traffic. Learn more.