dsh-book-reader 0.2.0 → 0.3.1
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/CHANGELOG.md +46 -2
- package/README.md +64 -28
- package/README.zh-CN.md +62 -26
- package/SKILL.md +22 -4
- package/lib/client.js +636 -217
- package/lib/core/README.md +47 -22
- package/lib/core/pdf.js +564 -35
- package/lib/index.js +167 -25
- package/package.json +1 -1
package/CHANGELOG.md
CHANGED
|
@@ -4,6 +4,47 @@ All notable changes to this project are documented here. The format follows
|
|
|
4
4
|
[Keep a Changelog](https://keepachangelog.com/en/1.1.0/), and this project
|
|
5
5
|
adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
|
|
6
6
|
|
|
7
|
+
## [0.3.0] — 2026-10-20
|
|
8
|
+
|
|
9
|
+
### Changed
|
|
10
|
+
|
|
11
|
+
- Search is now a basic in-page find bar: `Cmd/Ctrl+F` opens it, matches are
|
|
12
|
+
highlighted on the PDF text layer, and `Enter` / `Shift+Enter` (or the
|
|
13
|
+
previous/next buttons) jump between occurrences with an `n / total` counter.
|
|
14
|
+
- The search result list, its snippets, and the per-result “→ AI” insertion are
|
|
15
|
+
gone; search no longer inserts anything into the composer.
|
|
16
|
+
- `pdf_search` is now normalized text search: wrapping quotes/backticks and
|
|
17
|
+
`$…$` are stripped, NFKC folds math alphanumerics and ligatures, dashes and
|
|
18
|
+
line-break hyphenation are removed, and if no literal phrase matches it falls
|
|
19
|
+
back to a space-insensitive pass (`compact`) and then keyword AND (`terms`).
|
|
20
|
+
The result gains `mode`; hits rank body pages before front matter (at most 3
|
|
21
|
+
front-matter hits), `pages` counts every matched page, and the tool default
|
|
22
|
+
limit is 30. The tool description now states these rules so the model stops
|
|
23
|
+
passing quoted or LaTeX-marked phrases.
|
|
24
|
+
- `pdf_search` accepts an optional `pages` selection (the same syntax as
|
|
25
|
+
`pdf_read_pages`; printed pages by default, `page_numbers:"pdf"` for physical
|
|
26
|
+
pages) to scan only a page range and report matches inside it. This keeps a
|
|
27
|
+
late-chapter search from being crowded out by earlier pages.
|
|
28
|
+
|
|
29
|
+
### Fixed
|
|
30
|
+
|
|
31
|
+
- Opening a book now jumps to the remembered page immediately, before a saved
|
|
32
|
+
fixed zoom is replayed, so the cover no longer flashes for a few hundred
|
|
33
|
+
milliseconds while the zoom settles. The scroll is re-applied after the zoom
|
|
34
|
+
commit so the page stays aligned.
|
|
35
|
+
- `pdf_read_pages` no longer disappears from the model's tool list: its output
|
|
36
|
+
schema used a property-level `required: false`, which DSH's `defineTool`
|
|
37
|
+
rejects at registration time. The exception only reached the Host log, so the
|
|
38
|
+
plugin looked enabled with two tools instead of three.
|
|
39
|
+
|
|
40
|
+
## [0.3.1] — 2026-10-22
|
|
41
|
+
|
|
42
|
+
### Fixed
|
|
43
|
+
|
|
44
|
+
- Text-only `pdf_read_pages` calls (`include_image:false`) no longer crash while
|
|
45
|
+
rendering their output, and they no longer require an image-capable model
|
|
46
|
+
route or the attachment store; image reads still require both.
|
|
47
|
+
|
|
7
48
|
## [0.2.0] — 2026-10-20
|
|
8
49
|
|
|
9
50
|
### Added
|
|
@@ -16,8 +57,9 @@ adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
|
|
|
16
57
|
- New config key `rememberView` (default `true`).
|
|
17
58
|
- Host routes `GET /dsh-pdf-reader/view?path=…` and
|
|
18
59
|
`POST /dsh-pdf-reader/view` for the browser half.
|
|
19
|
-
- Docs: the README is split into
|
|
20
|
-
|
|
60
|
+
- Docs: the README is split into an English `README.md` and a Simplified Chinese
|
|
61
|
+
`README.zh-CN.md`, both rewritten in a product-page style (badges, quick start,
|
|
62
|
+
feature table, model-tool examples, FAQ, data/privacy notes).
|
|
21
63
|
|
|
22
64
|
## [0.1.1] — 2026-10-18
|
|
23
65
|
|
|
@@ -71,6 +113,8 @@ First public release.
|
|
|
71
113
|
- `pdfjs-dist` is pinned to `6.3.289` to match the host process.
|
|
72
114
|
- MIT licensed.
|
|
73
115
|
|
|
116
|
+
[0.3.1]: https://github.com/Strepsiades2/dsh-book-reader/releases/tag/v0.3.1
|
|
117
|
+
[0.3.0]: https://github.com/Strepsiades2/dsh-book-reader/releases/tag/v0.3.0
|
|
74
118
|
[0.2.0]: https://github.com/Strepsiades2/dsh-book-reader/releases/tag/v0.2.0
|
|
75
119
|
[0.1.1]: https://github.com/Strepsiades2/dsh-book-reader/releases/tag/v0.1.1
|
|
76
120
|
[0.1.0]: https://github.com/Strepsiades2/dsh-book-reader/releases/tag/v0.1.0
|
package/README.md
CHANGED
|
@@ -1,54 +1,90 @@
|
|
|
1
1
|
# dsh-book-reader
|
|
2
2
|
|
|
3
|
-
|
|
3
|
+
[](https://www.npmjs.com/package/dsh-book-reader)
|
|
4
|
+
[](LICENSE)
|
|
5
|
+
[](https://github.com/deepseek-ai/deepseek-harness)
|
|
6
|
+
|
|
7
|
+
**Read PDFs in DSH together with the model.** A native toolbar on the official PDF preview, plus three tools the model can use to look up the TOC, search the full text, and read pages — page images included, when a formula or diagram needs to be seen.
|
|
4
8
|
|
|
5
9
|
[中文说明](README.zh-CN.md)
|
|
6
10
|
|
|
7
|
-
##
|
|
11
|
+
## Why this one
|
|
12
|
+
|
|
13
|
+
- **Native, not a replacement.** It decorates DSH's own PDF preview, so zoom, text selection, reload, and Office→PDF keep working. No extra viewer, no sidebar dependency.
|
|
14
|
+
- **The model can actually read the book.** Three tools give it the page map, text search, and page content, with page images only when needed.
|
|
15
|
+
- **It remembers where you were.** Reopening a book restores the last page and zoom.
|
|
16
|
+
|
|
17
|
+
## Quick start
|
|
8
18
|
|
|
9
|
-
|
|
19
|
+
1. DSH **Plugins → Add plugin** → `dsh-book-reader` (or `dsh plugin add dsh-book-reader`). Restart DSH.
|
|
20
|
+
2. Open any `.pdf` in the right sidebar — the toolbar appears in the preview header.
|
|
21
|
+
3. Ask the model about the PDF; it can search and read from now on.
|
|
22
|
+
|
|
23
|
+
Optional — give the model the reading guide (costs, page numbering, pitfalls):
|
|
24
|
+
|
|
25
|
+
```sh
|
|
26
|
+
ln -sfn "<installed package dir>" "<workspace>/.dsh/skills/pdf-reader"
|
|
27
|
+
```
|
|
28
|
+
|
|
29
|
+
## Features
|
|
10
30
|
|
|
11
|
-
|
|
12
|
-
|
|
13
|
-
|
|
14
|
-
|
|
15
|
-
|
|
31
|
+
| | Feature | What it does |
|
|
32
|
+
| --- | --- | --- |
|
|
33
|
+
| 🧭 | Navigation | Previous/next page, page readout, jump box (printed page, or `p152` for PDF page 152) |
|
|
34
|
+
| 🔎 | Search | Full-text search (`Cmd/Ctrl+F`) with in-page highlights; `Enter` / `Shift+Enter` jump between occurrences, no snippets |
|
|
35
|
+
| 🤖 | AI position hint | One short “you are reading page N” line per turn, only when the page changed (`AI●` toggle) |
|
|
36
|
+
| 💾 | Reading memory | Remembers page and zoom per book; restores them when you reopen it |
|
|
37
|
+
| 💬 | Tooltips | Hover/focus explanations on every control, English or Chinese |
|
|
38
|
+
| 🧰 | Model tools | `pdf_info`, `pdf_search`, `pdf_read_pages` — see below |
|
|
16
39
|
|
|
17
|
-
|
|
40
|
+
## Model tools
|
|
18
41
|
|
|
19
|
-
| Tool |
|
|
20
|
-
| --- | --- |
|
|
21
|
-
| `pdf_info` | Page count, printed↔physical page offset, outline/TOC |
|
|
22
|
-
| `pdf_search` |
|
|
23
|
-
| `pdf_read_pages` |
|
|
42
|
+
| Tool | Use it for | Cost |
|
|
43
|
+
| --- | --- | --- |
|
|
44
|
+
| `pdf_info` | Page count, printed↔physical page offset, outline/TOC | text, ~free |
|
|
45
|
+
| `pdf_search` | Finding a phrase, theorem number, or figure caption. Tolerates quotes/LaTeX, hyphen and math-notation splits, with keyword fallbacks; body pages first; optional `pages` range | text, ~free |
|
|
46
|
+
| `pdf_read_pages` | Reading chosen pages as text, optionally with page images | text ~free; images ~350–1024 tokens/page |
|
|
24
47
|
|
|
25
|
-
|
|
48
|
+
Example prompts:
|
|
26
49
|
|
|
27
|
-
|
|
28
|
-
|
|
29
|
-
|
|
50
|
+
> “What does chapter 5 of this book cover?” → `pdf_info` + `pdf_read_pages`
|
|
51
|
+
> “Find where ‘singular value’ appears.” → `pdf_search`
|
|
52
|
+
> “Search only chapter 7 (printed pp. 270–295) for ‘SVD’.” → `pdf_search` with `pages:"270-295"`
|
|
53
|
+
> “Show me the formula in theorem 5.6.” → `pdf_read_pages` with images
|
|
30
54
|
|
|
31
|
-
|
|
32
|
-
ln -sfn "<installed package dir>" "<workspace>/.dsh/skills/pdf-reader"
|
|
33
|
-
```
|
|
55
|
+
The model decides when text is enough and when an image is worth it.
|
|
34
56
|
|
|
35
57
|
## Configuration
|
|
36
58
|
|
|
37
|
-
|
|
59
|
+
Every key is optional and the defaults are safe. Common ones:
|
|
38
60
|
|
|
39
61
|
- `rememberView` (default `true`) — remember page/zoom per book.
|
|
40
|
-
- `maxPagesPerCall` (default `4`) — page cap
|
|
41
|
-
- `defaultScale` (default `2.5`) — image render scale.
|
|
62
|
+
- `maxPagesPerCall` (default `4`) — page cap per read call.
|
|
63
|
+
- `defaultScale` (default `2.5`) — image render scale (~180 DPI).
|
|
42
64
|
- `autoShare` (default `true`) — the AI position hint.
|
|
43
65
|
|
|
44
66
|
The full list with defaults is in [`cordis.patch.yml`](cordis.patch.yml).
|
|
45
67
|
|
|
68
|
+
## FAQ
|
|
69
|
+
|
|
70
|
+
**The toolbar does not appear.** Only `.pdf` previews get it (not `.docx`/`.pptx`/images). Update/reinstall the plugin and restart DSH.
|
|
71
|
+
|
|
72
|
+
**The model does not know my page.** The hint is sent at most once per turn and only when the page changed; turn on `AI●`, or mention the page in your message.
|
|
73
|
+
|
|
74
|
+
**Scanned PDFs.** No text layer: search finds nothing and text reads are empty. Page images still work — check `textCoverage.isLikelyScanned` from `pdf_info`.
|
|
75
|
+
|
|
76
|
+
**Search finds nothing.** Quotes and `$…$` are stripped, hyphen/math splits and missing spaces have fallbacks, and when no literal phrase matches the tool falls back to keyword AND (`mode` in the result says which pass matched). Try one short distinctive word.
|
|
77
|
+
|
|
78
|
+
**Does it modify my PDF?** No. It only reads and renders; nothing is written next to your books.
|
|
79
|
+
|
|
80
|
+
**Where is the remembered state?** In DSH's data directory (`~/.dsh/storages/book_reader_views.json`), at most 500 books, keyed by file path.
|
|
81
|
+
|
|
82
|
+
**A DSH upgrade broke the toolbar.** The toolbar relies on the preview's private DOM markers; page tools keep working regardless.
|
|
83
|
+
|
|
46
84
|
## Notes
|
|
47
85
|
|
|
48
|
-
-
|
|
49
|
-
-
|
|
50
|
-
- Scanned PDFs have no text layer — page images are the only option (no OCR).
|
|
51
|
-
- `pdfjs-dist` is pinned to `6.3.289` to match the host process.
|
|
86
|
+
- Local HTTP routes are loopback-only and unauthenticated; model tools stay inside the session workspace by default (`allowOutsideWorkspace` opts out).
|
|
87
|
+
- `pdfjs-dist` is pinned to `6.3.289` to match the DSH host process.
|
|
52
88
|
|
|
53
89
|
## More
|
|
54
90
|
|
package/README.zh-CN.md
CHANGED
|
@@ -1,36 +1,58 @@
|
|
|
1
1
|
# dsh-book-reader
|
|
2
2
|
|
|
3
|
-
|
|
3
|
+
[](https://www.npmjs.com/package/dsh-book-reader)
|
|
4
|
+
[](LICENSE)
|
|
5
|
+
[](https://github.com/deepseek-ai/deepseek-harness)
|
|
6
|
+
|
|
7
|
+
**在 DSH 里和模型一起读 PDF。** 官方 PDF 预览上加一条原生工具栏，并给模型三个工具：查目录、全文搜索、读指定页——公式和图表需要时还能直接看页面图片。
|
|
4
8
|
|
|
5
9
|
[English](README.md)
|
|
6
10
|
|
|
7
|
-
##
|
|
11
|
+
## 为什么用它
|
|
12
|
+
|
|
13
|
+
- **原生，不替换。** 装饰 DSH 自己的 PDF 预览器，缩放、选中、刷新、Office→PDF 都保持原样；不额外装阅读器、不依赖侧边栏插件。
|
|
14
|
+
- **模型真的能读这本书。** 三个工具给它页码/目录、文本搜索和页面内容；只有需要看公式、图表时才花图片 token。
|
|
15
|
+
- **记住你读到哪。** 再次打开同一本书，自动回到上次的页码和缩放。
|
|
16
|
+
|
|
17
|
+
## 三步上手
|
|
8
18
|
|
|
9
|
-
|
|
19
|
+
1. DSH 的 **Plugins → Add plugin** → `dsh-book-reader`（或 `dsh plugin add dsh-book-reader`），重启 DSH。
|
|
20
|
+
2. 在右侧栏打开任意 `.pdf`，预览头部出现工具栏。
|
|
21
|
+
3. 直接问模型关于这本书的问题，它就能自己搜、自己读。
|
|
22
|
+
|
|
23
|
+
可选——把阅读指南交给模型（成本、页码规则、常见坑）：
|
|
24
|
+
|
|
25
|
+
```sh
|
|
26
|
+
ln -sfn "<已安装的包目录>" "<workspace>/.dsh/skills/pdf-reader"
|
|
27
|
+
```
|
|
28
|
+
|
|
29
|
+
## 功能
|
|
10
30
|
|
|
11
|
-
|
|
12
|
-
|
|
13
|
-
|
|
14
|
-
|
|
15
|
-
|
|
31
|
+
| | 功能 | 说明 |
|
|
32
|
+
| --- | --- | --- |
|
|
33
|
+
| 🧭 | 翻页与跳页 | 上一页/下一页、跟随滚动的页码读数、跳页框（印刷页码，或 `p152` 表示 PDF 第 152 页） |
|
|
34
|
+
| 🔎 | 全文搜索 | `Cmd/Ctrl+F` 打开搜索，命中文字在页面上高亮；`Enter` / `Shift+Enter` 在命中之间顺序跳转，不显示摘要 |
|
|
35
|
+
| 🤖 | AI 位置提示 | 每轮对话最多一行“你正在看第 N 页”，页码变了才发（`AI●` 开关） |
|
|
36
|
+
| 💾 | 阅读记忆 | 按书记住页码和缩放，重新打开时恢复 |
|
|
37
|
+
| 💬 | 悬停说明 | 每个控件都有悬停/聚焦提示，跟随 DSH 界面语言（中/英） |
|
|
38
|
+
| 🧰 | 模型工具 | `pdf_info`、`pdf_search`、`pdf_read_pages`，见下 |
|
|
16
39
|
|
|
17
|
-
|
|
40
|
+
## 模型工具
|
|
18
41
|
|
|
19
|
-
| 工具 |
|
|
20
|
-
| --- | --- |
|
|
21
|
-
| `pdf_info` | 页数、印刷页↔物理页偏移、目录/TOC |
|
|
22
|
-
| `pdf_search` |
|
|
23
|
-
| `pdf_read_pages` |
|
|
42
|
+
| 工具 | 用途 | 成本 |
|
|
43
|
+
| --- | --- | --- |
|
|
44
|
+
| `pdf_info` | 页数、印刷页↔物理页偏移、目录/TOC | 文本，几乎免费 |
|
|
45
|
+
| `pdf_search` | 找短语、定理编号、图表标题。容忍引号/LaTeX、连字符与数学符号拆分，支持关键词回退；正文页优先；可用 `pages` 限定页范围 | 文本，几乎免费 |
|
|
46
|
+
| `pdf_read_pages` | 读指定页的文本，可选返回页面图片 | 文本几乎免费；图片约 350–1024 token/页 |
|
|
24
47
|
|
|
25
|
-
|
|
48
|
+
可以这样问：
|
|
26
49
|
|
|
27
|
-
|
|
28
|
-
|
|
29
|
-
|
|
50
|
+
> “这本书第 5 章讲了什么？” → `pdf_info` + `pdf_read_pages`
|
|
51
|
+
> “找一下 singular value 出现在哪几页。” → `pdf_search`
|
|
52
|
+
> “只在第 7 章（印刷页 270–295）里找 SVD。” → `pdf_search` 加 `pages:"270-295"`
|
|
53
|
+
> “把定理 5.6 的公式截给我看。” → `pdf_read_pages` 带图片
|
|
30
54
|
|
|
31
|
-
|
|
32
|
-
ln -sfn "<已安装的包目录>" "<workspace>/.dsh/skills/pdf-reader"
|
|
33
|
-
```
|
|
55
|
+
什么时候用文本、什么时候值得花图片 token，由模型自己判断。
|
|
34
56
|
|
|
35
57
|
## 配置
|
|
36
58
|
|
|
@@ -38,17 +60,31 @@
|
|
|
38
60
|
|
|
39
61
|
- `rememberView`（默认 `true`）—— 按书记住页码/缩放。
|
|
40
62
|
- `maxPagesPerCall`（默认 `4`）—— 单次读页上限。
|
|
41
|
-
- `defaultScale`（默认 `2.5`）——
|
|
63
|
+
- `defaultScale`（默认 `2.5`）—— 图片渲染倍率（约 180 DPI）。
|
|
42
64
|
- `autoShare`（默认 `true`）—— AI 位置提示。
|
|
43
65
|
|
|
44
66
|
完整键和默认值见 [`cordis.patch.yml`](cordis.patch.yml) 注释。
|
|
45
67
|
|
|
68
|
+
## 常见问题
|
|
69
|
+
|
|
70
|
+
**看不到工具栏？** 只有 `.pdf` 预览会挂工具栏（`.docx`/`.pptx`/图片不会）。升级/重装插件后重启 DSH。
|
|
71
|
+
|
|
72
|
+
**模型不知道我在哪一页？** 位置提示每轮最多一次、页码变了才发；打开 `AI●`，或直接在消息里提页码。
|
|
73
|
+
|
|
74
|
+
**扫描件？** 没有文本层：搜索为空、文本读页为空；页面图片仍然可用，先用 `pdf_info` 的 `textCoverage.isLikelyScanned` 判断。
|
|
75
|
+
|
|
76
|
+
**搜索找不到东西？** 引号和 `$…$` 会被去掉，连字符/数学符号拆分、缺空格有回退；整句没有字面命中时会退化成关键词 AND 搜索（结果的 `mode` 会说明命中方式）。试一个更短、更独特的词。
|
|
77
|
+
|
|
78
|
+
**会改我的 PDF 吗？** 不会。只读取和渲染，不会往书目录写任何东西。
|
|
79
|
+
|
|
80
|
+
**记住的状态存哪？** DSH 数据目录下的 `~/.dsh/storages/book_reader_views.json`，按文件路径索引，最多保留 500 本。
|
|
81
|
+
|
|
82
|
+
**DSH 升级后工具栏失效？** 工具栏依赖预览器的私有 DOM 标记；即使失效，模型工具仍然可用。
|
|
83
|
+
|
|
46
84
|
## 说明
|
|
47
85
|
|
|
48
|
-
-
|
|
49
|
-
-
|
|
50
|
-
- 扫描件没有文本层，只能看图（没有 OCR）。
|
|
51
|
-
- `pdfjs-dist` 锁定 `6.3.289`，以和宿主进程一致。
|
|
86
|
+
- 本地 HTTP 路由只认回环地址、无鉴权；模型工具默认只在会话工作区内读取（可设 `allowOutsideWorkspace`）。
|
|
87
|
+
- `pdfjs-dist` 锁定 `6.3.289`，与 DSH 宿主进程保持一致。
|
|
52
88
|
|
|
53
89
|
## 更多
|
|
54
90
|
|
package/SKILL.md
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
---
|
|
2
2
|
name: pdf-reader
|
|
3
|
-
description: 'Read a PDF (book, paper, manual) cheaply and accurately with the dsh-book-reader tools: page map/TOC,
|
|
3
|
+
description: 'Read a PDF (book, paper, manual) cheaply and accurately with the dsh-book-reader tools: page map/TOC, normalized text search (quote/LaTeX-tolerant, with hyphen/space and keyword fallbacks), then selective page reads (text first, images only when typography matters). Use whenever the user asks about specific pages, a section, a theorem, a figure, or "this page" in a PDF.'
|
|
4
4
|
---
|
|
5
5
|
|
|
6
6
|
# Reading a PDF with `dsh-book-reader`
|
|
@@ -11,7 +11,7 @@ expense, not the first move.
|
|
|
11
11
|
| Tool | Cost | Use for |
|
|
12
12
|
| --- | --- | --- |
|
|
13
13
|
| `pdf_info` | ~0 (text) | page count, printed↔physical offset, outline/TOC |
|
|
14
|
-
| `pdf_search` | ~0 (text) | locating a phrase, theorem number, or figure caption |
|
|
14
|
+
| `pdf_search` | ~0 (text) | locating a phrase, theorem number, or figure caption (quotes and `$…$` are stripped; optional `pages` range) |
|
|
15
15
|
| `pdf_read_pages` + `include_image:false` | ~0 (text, per page) | prose, theorem statements, exercise text |
|
|
16
16
|
| `pdf_read_pages` (images on) | **~350–1024 tokens/page** | formulas, diagrams, tables, layout |
|
|
17
17
|
|
|
@@ -19,7 +19,12 @@ expense, not the first move.
|
|
|
19
19
|
|
|
20
20
|
1. `pdf_info` once per document → note `printedOffset` and the outline. Never
|
|
21
21
|
hardcode an offset; it is detected (`printed = pdf − printedOffset`).
|
|
22
|
-
2. `pdf_search` for a distinctive phrase → page numbers.
|
|
22
|
+
2. `pdf_search` for a short, distinctive phrase → page numbers. If the result's
|
|
23
|
+
`mode` is not `"phrase"` (i.e. `"compact"` or `"terms"`), treat the hits as
|
|
24
|
+
candidate pages and read them before quoting exact wording. When the reader
|
|
25
|
+
is late in the book (or the position hint says so), pass `pages:"…"` around
|
|
26
|
+
the current printed page (e.g. `pages:"270-295"`) so earlier chapters cannot
|
|
27
|
+
crowd the 30-hit budget; `page_numbers:"pdf"` switches to physical pages.
|
|
23
28
|
3. `pdf_read_pages` with **`include_image:false`** to read those pages as text.
|
|
24
29
|
4. Only if the text is ambiguous or the question is about typography (a formula,
|
|
25
30
|
a diagram, a table) → call again with images on, and crop.
|
|
@@ -82,7 +87,20 @@ needs it, and prefer `include_image:false` first.
|
|
|
82
87
|
- `pdf_info` also reports `title`, `author`, `fileSizeBytes`, `pageSize` and
|
|
83
88
|
`textCoverage` (`isLikelyScanned`). Check `textCoverage.isLikelyScanned`
|
|
84
89
|
*before* searching: a scanned PDF has no text layer, so go straight to images.
|
|
85
|
-
-
|
|
90
|
+
- **`pdf_search` is normalized text search, not a regex engine.** Wrapping
|
|
91
|
+
quotes/backticks and `$…$` are stripped; NFKC folds math alphanumerics
|
|
92
|
+
(`ℒ (𝑉)`→`L (V)`) and ligatures; dash variants are dropped, so line-break
|
|
93
|
+
hyphenation (`com- plex`) matches `complex`. If no literal phrase matches, the
|
|
94
|
+
tool falls back to a space-insensitive pass (`mode: "compact"`, so `L(V)`
|
|
95
|
+
finds `ℒ (𝑉)`) and then to keyword AND (`mode: "terms"`). Body pages rank
|
|
96
|
+
before front matter (at most 3 front-matter hits), `pages` counts every matched
|
|
97
|
+
page before `limit`, and the default limit is 30. If a query finds nothing, try
|
|
98
|
+
one shorter distinctive word and read the pages rather than assuming absence.
|
|
99
|
+
- **`pdf_search` can be scoped to a page range.** `pages:"270-295"` (same syntax
|
|
100
|
+
as `pdf_read_pages`, printed pages by default) only scans those pages; the
|
|
101
|
+
returned `pages` count then counts matches inside that range. Use it when the
|
|
102
|
+
user is reading a later chapter and whole-book hits would be dominated by
|
|
103
|
+
earlier pages.
|
|
86
104
|
- Page numbers are 1-based; relative paths resolve against the session directory.
|
|
87
105
|
- No annotation/note features: this plugin reads and renders only.
|
|
88
106
|
|