kyleandClaude Opus 5.5
cfde35397a
docs/plans/p5a: 01 kind+fetch limits, 02 pdftext child + runner, 03 dcache PDF tables, 04 pdfpool, 05 pdfdoc render (map/pages/grep), 06 api, 07 digest citations, metrics, CLI, skill, unit, README. Given tests compile against a panic-only skeleton and fail there for the intended reasons; PDFs for tests are built by internal/pdftest (given), plus two qpdf-encrypted fixtures. Owner changes for the run: vendor go-pdfkit/extract v0.3.0 + reader v0.6.0 (pinned via tools.go until task 02); AGENTS.md, tools/run-plan.sh (enforces a Co-Authored-By trailer for the model), docs/implementer-log.md, `make gate`; gofmt (whitespace) on clean.go and guard; TestUnsupportedType uses a zip now that PDFs are supported; requestDeadline 90→210 s and writeTimeout 120→240 s for the PDF path's worst case. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
87 lines
2.0 KiB
Go
87 lines
2.0 KiB
Go
package reader
|
|
|
|
import "fmt"
|
|
|
|
// LZW code points that are not table entries.
|
|
const (
|
|
lzwClear = 256
|
|
lzwEOD = 257
|
|
lzwFirst = 258
|
|
lzwMax = 4096
|
|
)
|
|
|
|
// lzwDecode expands PDF's variable-code-width LZW.
|
|
//
|
|
// The standard library's compress/lzw cannot be used here: PDF's /EarlyChange
|
|
// parameter defaults to 1, which widens the code one entry before the table
|
|
// actually fills, and compress/lzw implements neither that convention nor a
|
|
// way to ask for it. Getting this wrong does not fail — it silently produces
|
|
// plausible rubbish a few hundred bytes in.
|
|
func lzwDecode(data []byte, early bool) ([]byte, error) {
|
|
var table [lzwMax][]byte
|
|
for i := 0; i < 256; i++ {
|
|
table[i] = []byte{byte(i)}
|
|
}
|
|
next := lzwFirst
|
|
width := 9
|
|
bit := 0
|
|
var out, prev []byte
|
|
|
|
for {
|
|
if bit+width > len(data)*8 {
|
|
// A truncated stream keeps what it managed to say.
|
|
return out, nil
|
|
}
|
|
code := 0
|
|
for k := 0; k < width; k++ {
|
|
code = code<<1 | int(data[(bit+k)/8]>>(7-(bit+k)%8)&1)
|
|
|
|
}
|
|
bit += width
|
|
|
|
switch code {
|
|
case lzwEOD:
|
|
return out, nil
|
|
case lzwClear:
|
|
next, width, prev = lzwFirst, 9, nil
|
|
continue
|
|
}
|
|
|
|
var entry []byte
|
|
switch {
|
|
case code < next && table[code] != nil:
|
|
entry = table[code]
|
|
case code == next && prev != nil:
|
|
// The encoder may name the entry it is about to define.
|
|
entry = append(append([]byte{}, prev...), prev[0])
|
|
default:
|
|
// What the stream did say is kept, with the reason it stopped.
|
|
return out, fmt.Errorf("reader: LZWDecode: code %d is not in the table", code)
|
|
}
|
|
out = append(out, entry...)
|
|
|
|
if prev != nil && next < lzwMax {
|
|
table[next] = append(append([]byte{}, prev...), entry[0])
|
|
next++
|
|
}
|
|
prev = entry
|
|
|
|
// With EarlyChange the width grows one entry sooner. Both conventions
|
|
// are the same comparison against a shifted count.
|
|
count := next
|
|
if early {
|
|
count++
|
|
}
|
|
switch {
|
|
case count >= 2048:
|
|
width = 12
|
|
case count >= 1024:
|
|
width = 11
|
|
case count >= 512:
|
|
width = 10
|
|
default:
|
|
width = 9
|
|
}
|
|
}
|
|
}
|