docrev 0.9.18 → 0.10.0

package/lib/macro-filter.lua

@@ -0,0 +1,201 @@
+--[[
+docrev macro filter.
+
+Reads a JSON sidecar describing one-argument LaTeX-style macros and expands
+them per output FORMAT. Used for the built-in \tofill{X} (bold orange [X]
+placeholder) and any user-declared macros from rev.yaml.
+
+Sidecar path is passed via the DOCREV_MACROS_FILE environment variable, set
+by build.ts before spawning pandoc. Env vars (not metadata) because pandoc's
+filter traversal runs RawInline/RawBlock BEFORE Meta, so by the time we'd
+read metadata the inline expansions have already happened.
+
+Why raw OpenXML for docx? Pandoc 3.x's docx writer does NOT honor
+`Span{style="color: #..."}` — those spans render as plain text with no
+<w:color> run property. So for docx we emit raw <w:r> nodes directly. Same
+reasoning for the pptx-color-filter.
+
+For latex/pdf/beamer the markdown source already contains \tofill{X} as a raw
+LaTeX inline; we leave it alone because build.ts injects a \providecommand
+into header-includes. For html we emit a raw <span> with inline style. For
+everything else (markdown, gfm, plain) we degrade to **bold [X]** so the
+placeholder never silently disappears.
+]]
+
+local json = require('pandoc.json')
+
+local macros_by_name = {}
+
+local function load_sidecar()
+  local path = os.getenv('DOCREV_MACROS_FILE')
+  if not path or path == '' then
+    return
+  end
+  local fh = io.open(path, 'r')
+  if not fh then
+    io.stderr:write('docrev macro-filter: cannot read sidecar: ' .. path .. '\n')
+    return
+  end
+  local content = fh:read('*a')
+  fh:close()
+  local ok, parsed = pcall(json.decode, content)
+  if not ok or type(parsed) ~= 'table' or type(parsed.macros) ~= 'table' then
+    io.stderr:write('docrev macro-filter: malformed sidecar JSON\n')
+    return
+  end
+  for _, m in ipairs(parsed.macros) do
+    if type(m) == 'table' and type(m.name) == 'string' then
+      macros_by_name[m.name] = m
+    end
+  end
+end
+
+load_sidecar()
+
+local function xml_escape(s)
+  return (s:gsub('&', '&amp;'):gsub('<', '&lt;'):gsub('>', '&gt;'))
+end
+
+local function html_escape(s)
+  return (s
+    :gsub('&', '&amp;')
+    :gsub('<', '&lt;')
+    :gsub('>', '&gt;')
+    :gsub('"', '&quot;'))
+end
+
+-- Resolve effective style for a macro in the current pandoc format.
+-- Per-format entry wins over `default` (replacement, not merge — matches
+-- macros.ts semantics).
+local function pick_style(macro, format)
+  if macro.formats and macro.formats[format] then
+    return macro.formats[format]
+  end
+  return macro.default or {}
+end
+
+-- Build the inside of the bracket: [prefix][arg][suffix], optionally without
+-- brackets when style.bracket == false.
+local function compose_text(style, arg)
+  local prefix = style.prefix or ''
+  local suffix = style.suffix or ''
+  local inner = prefix .. arg .. suffix
+  if style.bracket == false then
+    return inner
+  end
+  return '[' .. inner .. ']'
+end
+
+local function render_docx_run(style, arg)
+  local rpr = {}
+  if style.color then
+    table.insert(rpr, '<w:color w:val="' .. style.color .. '"/>')
+  end
+  if style.bold then
+    table.insert(rpr, '<w:b/>')
+  end
+  if style.italic then
+    table.insert(rpr, '<w:i/>')
+  end
+  local rpr_xml = ''
+  if #rpr > 0 then
+    rpr_xml = '<w:rPr>' .. table.concat(rpr) .. '</w:rPr>'
+  end
+  local text = xml_escape(compose_text(style, arg))
+  return '<w:r>' .. rpr_xml ..
+         '<w:t xml:space="preserve">' .. text .. '</w:t></w:r>'
+end
+
+local function render_html(style, arg)
+  local css = {}
+  if style.color then
+    table.insert(css, 'color:#' .. style.color)
+  end
+  if style.bold then
+    table.insert(css, 'font-weight:bold')
+  end
+  if style.italic then
+    table.insert(css, 'font-style:italic')
+  end
+  local text = html_escape(compose_text(style, arg))
+  if #css == 0 then
+    return '<span>' .. text .. '</span>'
+  end
+  return '<span style="' .. table.concat(css, ';') .. '">' .. text .. '</span>'
+end
+
+-- Fallback path: produce native pandoc inlines so the macro never silently
+-- disappears in markdown/gfm/plain output. Used when the current format has
+-- no native rich-text path (or we couldn't open the sidecar).
+local function fallback_inlines(style, arg)
+  local doc = pandoc.read(compose_text(style, arg), 'markdown')
+  local inlines = pandoc.utils.blocks_to_inlines(doc.blocks)
+  if style.bold then
+    inlines = { pandoc.Strong(inlines) }
+  end
+  if style.italic then
+    inlines = { pandoc.Emph(inlines) }
+  end
+  return inlines
+end
+
+-- Match `\NAME{...}` (with balanced braces inside the argument is NOT
+-- supported — the use case is plain placeholder text, mirroring the reference
+-- filter; users who need nested braces should use a different mechanism).
+local function parse_call(text)
+  local name, arg = text:match('^\\([A-Za-z][A-Za-z0-9]*)%s*{(.*)}%s*$')
+  if name and arg and macros_by_name[name] then
+    return name, arg
+  end
+  return nil, nil
+end
+
+local function expand_inline(el)
+  if el.format ~= 'tex' and el.format ~= 'latex' then
+    return nil
+  end
+  local name, arg = parse_call(el.text)
+  if not name then return nil end
+  local macro = macros_by_name[name]
+  local style = pick_style(macro, FORMAT)
+
+  if FORMAT == 'docx' then
+    return pandoc.RawInline('openxml', render_docx_run(style, arg))
+  elseif FORMAT == 'html' or FORMAT == 'html4' or FORMAT == 'html5' or FORMAT == 'chunkedhtml' then
+    return pandoc.RawInline('html', render_html(style, arg))
+  elseif FORMAT == 'latex' or FORMAT == 'beamer' or FORMAT == 'context' then
+    -- Leave the raw LaTeX as-is. build.ts injects \providecommand into
+    -- header-includes, so the LaTeX engine renders it directly.
+    return nil
+  else
+    return fallback_inlines(style, arg)
+  end
+end
+
+local function expand_block(el)
+  if el.format ~= 'tex' and el.format ~= 'latex' then
+    return nil
+  end
+  local name, arg = parse_call(el.text)
+  if not name then return nil end
+  local macro = macros_by_name[name]
+  local style = pick_style(macro, FORMAT)
+
+  if FORMAT == 'docx' then
+    return pandoc.RawBlock('openxml', '<w:p>' .. render_docx_run(style, arg) .. '</w:p>')
+  elseif FORMAT == 'html' or FORMAT == 'html4' or FORMAT == 'html5' or FORMAT == 'chunkedhtml' then
+    return pandoc.RawBlock('html', '<p>' .. render_html(style, arg) .. '</p>')
+  elseif FORMAT == 'latex' or FORMAT == 'beamer' or FORMAT == 'context' then
+    return nil
+  else
+    return pandoc.Para(fallback_inlines(style, arg))
+  end
+end
+
+function RawInline(el)
+  return expand_inline(el)
+end
+
+function RawBlock(el)
+  return expand_block(el)
+end

package/lib/macros.ts

@@ -0,0 +1,273 @@
+/**
+ * Placeholder/highlight macros for docrev.
+ *
+ * Users write `\tofill{X}` (or any custom macro they declare) in markdown
+ * source and the build pipeline expands it per output format:
+ *
+ *   - docx: raw OpenXML run with explicit color + bold (Span+style is NOT
+ *     honored by pandoc's docx writer, so we emit raw <w:r> nodes).
+ *   - pdf / tex / beamer: a `\providecommand` is injected via header-includes,
+ *     so the LaTeX command works directly. `\providecommand` means the user
+ *     can still override with `\renewcommand` in their own preamble.
+ *   - html: raw HTML span with inline style.
+ *   - everything else (markdown, gfm, etc.): bold [X] fallback. Never silently
+ *     dropped.
+ *
+ * The mechanism is generic: `\tofill` is the first built-in. Users can add
+ * their own macros under `macros:` in rev.yaml, and override the built-in by
+ * declaring a macro with the same name.
+ */
+
+import * as fs from 'fs';
+import * as path from 'path';
+import { fileURLToPath } from 'url';
+
+// =============================================================================
+// Types
+// =============================================================================
+
+/**
+ * Per-format rendering rules for a macro.
+ *
+ * Fields are independent — set any subset. Unset fields fall back to defaults
+ * (no color, no bold, no italic, bracket wrap on, etc.).
+ */
+export interface MacroFormatStyle {
+  /** Hex color without '#' (e.g. "C2410C"). */
+  color?: string;
+  /** Wrap the rendered content in bold. */
+  bold?: boolean;
+  /** Wrap the rendered content in italic. */
+  italic?: boolean;
+  /** Wrap the content in [...] brackets. Default: true. */
+  bracket?: boolean;
+  /** Optional literal prefix string inside the brackets (e.g. "NOTE: "). */
+  prefix?: string;
+  /** Optional literal suffix string inside the brackets. */
+  suffix?: string;
+}
+
+/**
+ * Macro definition. `name` is the LaTeX command name without the leading
+ * backslash (e.g. "tofill" → \tofill{...}). `formats` holds per-format rules;
+ * a missing format key inherits from `default`.
+ */
+export interface MacroDef {
+  name: string;
+  /** Default rendering rules; used when a format-specific override is absent. */
+  default?: MacroFormatStyle;
+  /** Per-format overrides, keyed by pandoc format (docx, pdf, html, ...). */
+  formats?: Record<string, MacroFormatStyle>;
+}
+
+/**
+ * Built-in macros shipped with docrev. The first entry is the original use
+ * case: \tofill{X} → bold orange [X] placeholder.
+ */
+export const BUILTIN_MACROS: MacroDef[] = [
+  {
+    name: 'tofill',
+    default: { color: 'C2410C', bold: true, bracket: true },
+  },
+];
+
+// =============================================================================
+// Validation
+// =============================================================================
+
+const MACRO_NAME_RE = /^[A-Za-z][A-Za-z0-9]*$/;
+const HEX_COLOR_RE = /^[0-9A-Fa-f]{6}$/;
+
+/**
+ * Validate a user-declared macro entry. Returns a list of error strings;
+ * empty means the macro is valid.
+ */
+export function validateMacro(macro: unknown): string[] {
+  const errors: string[] = [];
+
+  if (!macro || typeof macro !== 'object' || Array.isArray(macro)) {
+    return ['macro must be an object'];
+  }
+  const m = macro as Partial<MacroDef>;
+
+  if (!m.name || typeof m.name !== 'string') {
+    errors.push('macro.name is required');
+  } else if (!MACRO_NAME_RE.test(m.name)) {
+    errors.push(`macro.name "${m.name}" must match [A-Za-z][A-Za-z0-9]*`);
+  }
+
+  const checkStyle = (style: unknown, key: string): void => {
+    if (style === undefined) return;
+    if (!style || typeof style !== 'object' || Array.isArray(style)) {
+      errors.push(`${key} must be an object`);
+      return;
+    }
+    const s = style as MacroFormatStyle;
+    if (s.color !== undefined && (typeof s.color !== 'string' || !HEX_COLOR_RE.test(s.color))) {
+      errors.push(`${key}.color must be a 6-digit hex string without '#' (got "${s.color}")`);
+    }
+    for (const flag of ['bold', 'italic', 'bracket'] as const) {
+      if (s[flag] !== undefined && typeof s[flag] !== 'boolean') {
+        errors.push(`${key}.${flag} must be a boolean`);
+      }
+    }
+    for (const text of ['prefix', 'suffix'] as const) {
+      if (s[text] !== undefined && typeof s[text] !== 'string') {
+        errors.push(`${key}.${text} must be a string`);
+      }
+    }
+  };
+
+  checkStyle(m.default, 'macro.default');
+  if (m.formats !== undefined) {
+    if (!m.formats || typeof m.formats !== 'object' || Array.isArray(m.formats)) {
+      errors.push('macro.formats must be an object keyed by format name');
+    } else {
+      for (const [fmt, style] of Object.entries(m.formats)) {
+        checkStyle(style, `macro.formats.${fmt}`);
+      }
+    }
+  }
+
+  return errors;
+}
+
+// =============================================================================
+// Merge
+// =============================================================================
+
+/**
+ * Merge built-in macros with user-declared macros. User entries override
+ * built-ins by `name` (case-sensitive). Invalid entries are dropped with a
+ * console warning so a malformed user macro never silently disables the
+ * built-in.
+ */
+export function mergeMacros(userMacros: unknown): MacroDef[] {
+  const builtins = new Map<string, MacroDef>();
+  for (const m of BUILTIN_MACROS) builtins.set(m.name, m);
+
+  if (!userMacros) return [...builtins.values()];
+  if (!Array.isArray(userMacros)) {
+    console.warn('macros: rev.yaml `macros` must be a list; ignoring');
+    return [...builtins.values()];
+  }
+
+  for (const raw of userMacros) {
+    const errors = validateMacro(raw);
+    if (errors.length > 0) {
+      console.warn(`macros: skipping invalid macro: ${errors.join('; ')}`);
+      continue;
+    }
+    const def = raw as MacroDef;
+    builtins.set(def.name, def);
+  }
+
+  return [...builtins.values()];
+}
+
+// =============================================================================
+// LaTeX preamble generation (PDF / tex / beamer)
+// =============================================================================
+
+/**
+ * Build the LaTeX color spec for a style. Returns the wrapping LaTeX with a
+ * `#1` placeholder where the argument lands.
+ */
+function latexCommandBody(style: MacroFormatStyle): string {
+  const prefix = style.prefix ? escapeLatex(style.prefix) : '';
+  const suffix = style.suffix ? escapeLatex(style.suffix) : '';
+  const inner = `${prefix}#1${suffix}`;
+  const bracketed = style.bracket === false ? inner : `[${inner}]`;
+
+  let body = bracketed;
+  if (style.italic) body = `\\textit{${body}}`;
+  if (style.bold) body = `\\textbf{${body}}`;
+  if (style.color) body = `\\textcolor[HTML]{${style.color.toUpperCase()}}{${body}}`;
+  return body;
+}
+
+/**
+ * Generate `\providecommand` definitions for all macros. `\providecommand`
+ * means user-supplied `\renewcommand` (in a custom header-includes file) still
+ * wins, preserving backwards compat with existing projects.
+ *
+ * Returns an empty string when the macro list is empty.
+ */
+export function generateLatexPreamble(macros: MacroDef[]): string {
+  const lines: string[] = ['% docrev: placeholder macros'];
+  // \textcolor in [HTML]{...} requires xcolor with the [HTML] option.
+  lines.push('\\PassOptionsToPackage{HTML}{xcolor}');
+  // Some templates already load xcolor; \usepackage tolerates duplicates with
+  // the same options.
+  lines.push('\\usepackage[HTML]{xcolor}');
+
+  for (const m of macros) {
+    const style = pickStyle(m, 'latex');
+    const body = latexCommandBody(style);
+    lines.push(`\\providecommand{\\${m.name}}[1]{${body}}`);
+  }
+
+  return lines.join('\n');
+}
+
+function escapeLatex(s: string): string {
+  // Conservative escape — these are user-authored short literals (prefix/suffix).
+  return s
+    .replace(/\\/g, '\\textbackslash{}')
+    .replace(/([&%$#_{}])/g, '\\$1')
+    .replace(/~/g, '\\textasciitilde{}')
+    .replace(/\^/g, '\\textasciicircum{}');
+}
+
+// =============================================================================
+// Style resolution per format
+// =============================================================================
+
+/**
+ * Resolve the effective style for a macro in a given pandoc format. Per-format
+ * override wins over the macro's default; both can be partial — fields are
+ * not merged across `default` and `formats[fmt]` (the format-specific entry
+ * replaces `default` entirely when present), keeping rev.yaml semantics
+ * predictable.
+ *
+ * Falls back to `default` when no `formats[fmt]` exists, and to an empty
+ * style ({}) when neither is set.
+ */
+export function pickStyle(macro: MacroDef, format: string): MacroFormatStyle {
+  const fmt = macro.formats?.[format];
+  if (fmt) return fmt;
+  return macro.default ?? {};
+}
+
+// =============================================================================
+// Lua filter sidecar
+// =============================================================================
+
+/**
+ * Serialize the macro list to a compact JSON sidecar consumed by the lua
+ * filter at build time. The lua filter reads this file at startup and uses it
+ * to expand `\tofill{X}` (or any other declared macro) per FORMAT.
+ *
+ * Returns the absolute path to the written sidecar.
+ */
+export function writeMacrosSidecar(directory: string, macros: MacroDef[]): string {
+  const sidecarPath = path.join(directory, '.macros.json');
+  fs.writeFileSync(sidecarPath, JSON.stringify({ macros }), 'utf-8');
+  return sidecarPath;
+}
+
+/**
+ * Resolve the absolute path to the bundled lua filter. Works both from source
+ * (`lib/macro-filter.lua`) and from the compiled package (`dist/lib/...`)
+ * because the postbuild script copies .lua files alongside the .js output.
+ */
+export function getMacroFilterPath(): string {
+  // import.meta.url points to the running file: lib/macros.ts in source,
+  // dist/lib/macros.js when published. The filter sits next to it.
+  //
+  // Use fileURLToPath so paths with spaces (Windows: "C:\Users\Gilles Colling\…")
+  // resolve correctly. The naive `new URL(...).pathname` returns URL-encoded
+  // `%20` segments and fs.existsSync silently fails.
+  const here = path.dirname(fileURLToPath(import.meta.url));
+  return path.join(here, 'macro-filter.lua');
+}

package/lib/schema.ts

@@ -218,6 +218,40 @@ export const revYamlSchema: Schema = {
       },
       additionalProperties: true,
     },
+    macros: {
+      type: 'array',
+      description: 'Placeholder/highlight macros (e.g. \\tofill{X}). Built-in macros are merged automatically; entries here add new macros or override built-ins by name.',
+      items: {
+        type: 'object',
+        properties: {
+          name: {
+            type: 'string',
+            description: 'Macro command name without leading backslash, e.g. "tofill" for \\tofill{...}',
+            pattern: '^[A-Za-z][A-Za-z0-9]*$',
+          },
+          default: {
+            type: 'object',
+            description: 'Default rendering rules across formats',
+            properties: {
+              color: { type: 'string', pattern: '^[0-9A-Fa-f]{6}$' },
+              bold: { type: 'boolean' },
+              italic: { type: 'boolean' },
+              bracket: { type: 'boolean', default: true },
+              prefix: { type: 'string' },
+              suffix: { type: 'string' },
+            },
+            additionalProperties: false,
+          },
+          formats: {
+            type: 'object',
+            description: 'Per-format overrides (keys: docx, pdf, latex, html, ...)',
+            additionalProperties: true,
+          },
+        },
+        required: ['name'],
+        additionalProperties: false,
+      },
+    },
   },
   additionalProperties: true,
 };

package/mkdocs.yml

@@ -1,64 +1,64 @@
-site_name: docrev
-site_url: https://gillescolling.com/docrev
-site_description: CLI for writing documents in Markdown while collaborating with Word users.
-site_author: Gilles Colling
-repo_url: https://github.com/gcol33/docrev
-repo_name: gcol33/docrev
-
-theme:
-  name: material
-  palette:
-    - scheme: default
-      primary: custom
-      accent: custom
-      toggle:
-        icon: material/brightness-7
-        name: Switch to dark mode
-    - scheme: slate
-      primary: custom
-      accent: custom
-      toggle:
-        icon: material/brightness-4
-        name: Switch to light mode
-  font:
-    text: Roboto
-    code: Roboto Mono
-  features:
-    - navigation.tabs
-    - navigation.top
-    - navigation.instant
-    - search.highlight
-    - content.code.copy
-  icon:
-    repo: fontawesome/brands/github
-
-nav:
-  - Home: index.md
-  - Get Started: workflow.md
-  - Commands: commands.md
-  - Configuration: configuration.md
-  - Troubleshooting: troubleshooting.md
-
-markdown_extensions:
-  - pymdownx.highlight:
-      anchor_linenums: true
-  - pymdownx.superfences
-  - pymdownx.inlinehilite
-  - pymdownx.tabbed:
-      alternate_style: true
-  - admonition
-  - pymdownx.details
-  - attr_list
-  - md_in_html
-  - toc:
-      permalink: true
-
-extra_css:
-  - stylesheets/extra.css
-
-extra:
-  social:
-    - icon: fontawesome/brands/github
-      link: https://github.com/gcol33/docrev
-    - icon: fontawesome/brands/npm
-      link: https://www.npmjs.com/package/docrev
+site_name: docrev
+site_url: https://gillescolling.com/docrev
+site_description: CLI for writing documents in Markdown while collaborating with Word users.
+site_author: Gilles Colling
+repo_url: https://github.com/gcol33/docrev
+repo_name: gcol33/docrev
+
+theme:
+  name: material
+  palette:
+    - scheme: default
+      primary: custom
+      accent: custom
+      toggle:
+        icon: material/brightness-7
+        name: Switch to dark mode
+    - scheme: slate
+      primary: custom
+      accent: custom
+      toggle:
+        icon: material/brightness-4
+        name: Switch to light mode
+  font:
+    text: Roboto
+    code: Roboto Mono
+  features:
+    - navigation.tabs
+    - navigation.top
+    - navigation.instant
+    - search.highlight
+    - content.code.copy
+  icon:
+    repo: fontawesome/brands/github
+
+nav:
+  - Home: index.md
+  - Get Started: workflow.md
+  - Commands: commands.md
+  - Configuration: configuration.md
+  - Troubleshooting: troubleshooting.md
+
+markdown_extensions:
+  - pymdownx.highlight:
+      anchor_linenums: true
+  - pymdownx.superfences
+  - pymdownx.inlinehilite
+  - pymdownx.tabbed:
+      alternate_style: true
+  - admonition
+  - pymdownx.details
+  - attr_list
+  - md_in_html
+  - toc:
+      permalink: true
+
+extra_css:
+  - stylesheets/extra.css
+
+extra:
+  social:
+    - icon: fontawesome/brands/github
+      link: https://github.com/gcol33/docrev
+    - icon: fontawesome/brands/npm
+      link: https://www.npmjs.com/package/docrev

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "docrev",
-  "version": "0.9.18",
+  "version": "0.10.0",
   "description": "Academic paper revision workflow: Word ↔ Markdown round-trips, DOI validation, reviewer comments",
   "type": "module",
   "types": "dist/lib/types.d.ts",

package/scripts/postbuild.js

@@ -6,14 +6,20 @@
  * tsc compiles bin/rev.ts → dist/bin/rev.js but:
  *   1. Preserves the #!/usr/bin/env tsx shebang (needs to be node)
  *   2. Relative paths like '../package.json' break (bin/ → dist/bin/ adds a level)
+ *
+ * Also copies non-TS asset files (lua filters) from lib/ to dist/lib/ so the
+ * compiled output can locate them via `import.meta.url`. Without this step
+ * the lua filters live in lib/ in the published tarball while the runtime
+ * looks for them in dist/lib/.
  */
 
-import { readFileSync, writeFileSync } from 'fs';
+import { readFileSync, writeFileSync, readdirSync, copyFileSync, mkdirSync, existsSync } from 'fs';
 import { join, dirname } from 'path';
 import { fileURLToPath } from 'url';
 
 const __dirname = dirname(fileURLToPath(import.meta.url));
-const revPath = join(__dirname, '..', 'dist', 'bin', 'rev.js');
+const projectRoot = join(__dirname, '..');
+const revPath = join(projectRoot, 'dist', 'bin', 'rev.js');
 
 let content = readFileSync(revPath, 'utf-8');
 
@@ -26,3 +32,16 @@ content = content.replace("'../package.json'", "'../../package.json'");
 
 writeFileSync(revPath, content, 'utf-8');
 console.log('postbuild: fixed dist/bin/rev.js (shebang + paths)');
+
+// Copy lua filter assets so import.meta.url resolves them at runtime.
+const libDir = join(projectRoot, 'lib');
+const distLibDir = join(projectRoot, 'dist', 'lib');
+if (!existsSync(distLibDir)) {
+  mkdirSync(distLibDir, { recursive: true });
+}
+for (const entry of readdirSync(libDir)) {
+  if (entry.endsWith('.lua')) {
+    copyFileSync(join(libDir, entry), join(distLibDir, entry));
+    console.log(`postbuild: copied ${entry} → dist/lib/`);
+  }
+}