PDF Reader (Superpower Fork)

Fork of github.com/ledongthuc/pdf with fixes for malformed PDF CMap handling.

Why This Fork Exists

The upstream library fails to extract text from PDFs that have malformed CMap (character mapping) definitions. Specifically, some PDF generators create CMaps where:

• The codespace range is defined too narrowly (e.g., <51> to <79>)
• But the bfrange mappings include characters outside that range (e.g., <20> for space, <30> for digit 0)

The upstream library's cmap.Decode() function only looks up character mappings if the byte falls within a declared codespace range. Characters outside the codespace are replaced with the Unicode replacement character, resulting in garbled text output.

In Simple Terms

Without the patch, extracting text from affected PDFs produces garbage:

��������� ������ - �����
���������� �����*:                    ��� ��/��

With the patch, you get the actual text:

Glyphosate Profile - Urine
Creatinine Value*:                    100 mg/dl

The fix allows the library to find character mappings even when the PDF's internal character map is technically malformed.

Changes From Upstream

File: page.go - Modified cmap.Decode() function

Added a fallback that tries bfchar/bfrange lookups even when bytes are outside the declared codespace. This handles malformed CMaps that define mappings outside their declared codespace range.

// After the codespace loop fails, try bfchar/bfrange lookups for single-byte
// codes even when outside the declared codespace. Some PDFs have malformed
// CMaps where the codespace range is too narrow but bfrange mappings exist.

Install

go get -u github.com/superpowerdotcom/go-pdf-lib

Usage

package main

import (
    "bytes"
    "fmt"

    pdf "github.com/superpowerdotcom/go-pdf-lib"
)

func main() {
    f, r, err := pdf.Open("./test.pdf")
    if err != nil {
        panic(err)
    }
    defer f.Close()

    var buf bytes.Buffer
    b, err := r.GetPlainText()
    if err != nil {
        panic(err)
    }
    buf.ReadFrom(b)
    fmt.Println(buf.String())
}

Read Text By Row (preserves layout)

package main

import (
    "fmt"

    pdf "github.com/superpowerdotcom/go-pdf-lib"
)

func main() {
    f, r, err := pdf.Open("./test.pdf")
    if err != nil {
        panic(err)
    }
    defer f.Close()

    p := r.Page(1)
    rows, err := p.GetTextByRow()
    if err != nil {
        panic(err)
    }

    for _, row := range rows {
        fmt.Printf("Y=%.1f: ", row.Content[0].Y)
        for _, word := range row.Content {
            fmt.Print(word.S, " ")
        }
        fmt.Println()
    }
}

Original Features

• Get plain text content (without format)
• Get Content (including all font and formatting information)
• Get text grouped by rows or columns

License

BSD-style license (see LICENSE file)