ctrl-kd 1.1.2__tar.gz

ctrl_kd-1.1.2/LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 Jon Michaels
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

ctrl_kd-1.1.2/PKG-INFO

@@ -0,0 +1,120 @@
+Metadata-Version: 2.4
+Name: ctrl-kd
+Version: 1.1.2
+Summary: Convert WordStar 4-7 documents and print-to-disk files to text, Markdown, HTML, RTF, or PDF. ^KD: save and done.
+Author: Jon Michaels
+License: MIT
+Project-URL: Homepage, https://github.com/jonmichaels/ctrl-kd
+Keywords: wordstar,converter,retrocomputing,archive,dos
+Classifier: Development Status :: 5 - Production/Stable
+Classifier: Environment :: Console
+Classifier: License :: OSI Approved :: MIT License
+Classifier: Programming Language :: Python :: 3
+Classifier: Topic :: Text Processing :: Filters
+Classifier: Topic :: System :: Archiving
+Requires-Python: >=3.9
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Dynamic: license-file
+
+# ctrl-kd
+
+Convert WordStar-era files to modern formats. **^KD: save and done.**
+
+`ctrl-kd` reads WordStar 4 documents, WordStar 5–7 documents, and WordStar
+**print-to-disk files** (the printer byte stream, captured to a file — a distinct
+format most converters mangle), and writes plain text, Markdown, HTML, RTF, or PDF (typewriter-set on the
+built-in Courier fonts — no dependencies, the page as it would have printed).
+
+```console
+$ ctrl-kd ESSAY.WS                      # -> ESSAY.md
+$ ctrl-kd ESSAY.WS -t html -t rtf       # multiple formats
+$ ctrl-kd ESSAY.WS -t pdf --mode printed # a facsimile of the 1990 printout
+$ ctrl-kd --mode printed LETTER.WS      # line-for-line, as it printed in 1990
+$ ctrl-kd --diagnose MYSTERY.FIL        # what IS this file?
+```
+
+## Why another converter?
+
+Existing tools each lose something. Fed a WordStar 4 file, converters written for
+WS7 delete the last letter of every word (WS4 set bit 7 on it). Most delete soft
+returns outright — `Jon Michaels` + `March 6, 1992` becomes
+`Jon MichaelsMarch 6, 1992` — which also destroys every poem, because poem lines
+end in soft returns too. And print-to-disk files aren't WordStar documents at all,
+so feeding them to a WordStar converter produces stray superscripts and garbage.
+
+`ctrl-kd` was built by converting a real 1987–1992 corpus (high-school and college
+papers, poems, stories — WordStar 4 on DOS, dot-matrix printer) and verifying
+against surviving period printouts of the same documents. Its rules are empirical:
+
+* **Detection by content, never by extension.** WS4 vs WS5+ vs print stream vs
+  plain text vs binary, with the evidence shown in `--diagnose`.
+* **The wrap test.** WordStar wrapped only when the next word didn't fit. So a
+  soft return where the next word *would* have fit (strictly — WordStar wrapped
+  even on an exact-margin fit) is a deliberate break: a poem line, a heading.
+  Everything else is word wrap and joins with a space. The margin is estimated
+  from the 90th percentile of soft-wrapped line lengths (floor 65, the default).
+* **Break runs.** Soft/hard return runs containing a hard return and a blank line
+  are paragraph breaks; a lone hard return is the author's deliberate line break.
+  Double-spaced documents (blank soft lines between every line) collapse
+  automatically.
+* **Ruler lines mean columns.** A `.rr----!----` dot line defines tab stops; the
+  document's alignment is space-built and only survives fixed-width. Such
+  documents render `printed` in every mode.
+* **Print streams render verbatim** — they ARE the printed page — with printer
+  style codes decoded (superscript/underline/italic/bold pairs; table in
+  `core.PRINT_CODES`, derived from a late-80s dot-matrix driver and overridable).
+* **WS5+ symmetric blocks** (`0x1D`: real footnotes/endnotes, headings, page
+  breaks — machinery added in WS5) are parsed with their nested structure,
+  verified against the 86 WordStar 7 documents in Robert J. Sawyer's public
+  WordStar archive: footnotes extract with in-text references (`[^n]` in
+  Markdown), paragraph styles become headings, and 82/86 convert with zero
+  mojibake. More WS5–7 corpora still welcome.
+
+## Modes
+
+* `--mode modern` (default): reflowed paragraphs, semantic markup, deliberate
+  line breaks kept.
+* `--mode printed`: every line as laid out, fixed-width, `.pa`/form-feed page
+  breaks honored — how it came off the printer.
+
+## Install
+
+Straight from GitHub (not yet on PyPI):
+
+```console
+$ pipx install git+https://github.com/jonmichaels/ctrl-kd
+```
+
+or `pip install git+https://github.com/jonmichaels/ctrl-kd` into an environment
+of your choice. Python ≥ 3.9, no dependencies.
+Library API: `ctrlkd.convert(data, to='html')`.
+
+## Adding an output format
+
+An output format is one function over the parsed document — register it with the
+`@ctrlkd.emitter` decorator, or ship it as a pip-installable plugin via the
+`ctrlkd.emitters` entry-point group and it appears in the CLI automatically.
+**[EXTENDING.md](EXTENDING.md)** has the IR contract, a complete worked example
+(BBCode in ~40 lines), and a checklist.
+
+## Lineage
+
+Standing on the shoulders of the tools and documentation that kept WordStar
+readable: Yohanes Nugroho's WS-CON, Michael Petrie's English port, the `wsconvert`
+project, Robert J. Sawyer's WordStar archive, and the WordStar format
+documentation community. Behaviors were studied and reimplemented; no code was
+copied. The development corpus is personal and is not distributed — tests use
+synthetic fixtures that encode the same behaviors.
+
+## Credits
+
+Written by Jon Michaels — whose 1987–1992 WordStar files, and the need to read
+them again, are the reason this exists — with Athena (Claude, Anthropic) as
+co-author: the byte archaeology, the wrap test, and the implementation grew out
+of a joint effort to recover those disks. Every commit carries the co-author
+trailer.
+
+## License
+
+MIT © Jon Michaels

ctrl_kd-1.1.2/README.md

@@ -0,0 +1,101 @@
+# ctrl-kd
+
+Convert WordStar-era files to modern formats. **^KD: save and done.**
+
+`ctrl-kd` reads WordStar 4 documents, WordStar 5–7 documents, and WordStar
+**print-to-disk files** (the printer byte stream, captured to a file — a distinct
+format most converters mangle), and writes plain text, Markdown, HTML, RTF, or PDF (typewriter-set on the
+built-in Courier fonts — no dependencies, the page as it would have printed).
+
+```console
+$ ctrl-kd ESSAY.WS                      # -> ESSAY.md
+$ ctrl-kd ESSAY.WS -t html -t rtf       # multiple formats
+$ ctrl-kd ESSAY.WS -t pdf --mode printed # a facsimile of the 1990 printout
+$ ctrl-kd --mode printed LETTER.WS      # line-for-line, as it printed in 1990
+$ ctrl-kd --diagnose MYSTERY.FIL        # what IS this file?
+```
+
+## Why another converter?
+
+Existing tools each lose something. Fed a WordStar 4 file, converters written for
+WS7 delete the last letter of every word (WS4 set bit 7 on it). Most delete soft
+returns outright — `Jon Michaels` + `March 6, 1992` becomes
+`Jon MichaelsMarch 6, 1992` — which also destroys every poem, because poem lines
+end in soft returns too. And print-to-disk files aren't WordStar documents at all,
+so feeding them to a WordStar converter produces stray superscripts and garbage.
+
+`ctrl-kd` was built by converting a real 1987–1992 corpus (high-school and college
+papers, poems, stories — WordStar 4 on DOS, dot-matrix printer) and verifying
+against surviving period printouts of the same documents. Its rules are empirical:
+
+* **Detection by content, never by extension.** WS4 vs WS5+ vs print stream vs
+  plain text vs binary, with the evidence shown in `--diagnose`.
+* **The wrap test.** WordStar wrapped only when the next word didn't fit. So a
+  soft return where the next word *would* have fit (strictly — WordStar wrapped
+  even on an exact-margin fit) is a deliberate break: a poem line, a heading.
+  Everything else is word wrap and joins with a space. The margin is estimated
+  from the 90th percentile of soft-wrapped line lengths (floor 65, the default).
+* **Break runs.** Soft/hard return runs containing a hard return and a blank line
+  are paragraph breaks; a lone hard return is the author's deliberate line break.
+  Double-spaced documents (blank soft lines between every line) collapse
+  automatically.
+* **Ruler lines mean columns.** A `.rr----!----` dot line defines tab stops; the
+  document's alignment is space-built and only survives fixed-width. Such
+  documents render `printed` in every mode.
+* **Print streams render verbatim** — they ARE the printed page — with printer
+  style codes decoded (superscript/underline/italic/bold pairs; table in
+  `core.PRINT_CODES`, derived from a late-80s dot-matrix driver and overridable).
+* **WS5+ symmetric blocks** (`0x1D`: real footnotes/endnotes, headings, page
+  breaks — machinery added in WS5) are parsed with their nested structure,
+  verified against the 86 WordStar 7 documents in Robert J. Sawyer's public
+  WordStar archive: footnotes extract with in-text references (`[^n]` in
+  Markdown), paragraph styles become headings, and 82/86 convert with zero
+  mojibake. More WS5–7 corpora still welcome.
+
+## Modes
+
+* `--mode modern` (default): reflowed paragraphs, semantic markup, deliberate
+  line breaks kept.
+* `--mode printed`: every line as laid out, fixed-width, `.pa`/form-feed page
+  breaks honored — how it came off the printer.
+
+## Install
+
+Straight from GitHub (not yet on PyPI):
+
+```console
+$ pipx install git+https://github.com/jonmichaels/ctrl-kd
+```
+
+or `pip install git+https://github.com/jonmichaels/ctrl-kd` into an environment
+of your choice. Python ≥ 3.9, no dependencies.
+Library API: `ctrlkd.convert(data, to='html')`.
+
+## Adding an output format
+
+An output format is one function over the parsed document — register it with the
+`@ctrlkd.emitter` decorator, or ship it as a pip-installable plugin via the
+`ctrlkd.emitters` entry-point group and it appears in the CLI automatically.
+**[EXTENDING.md](EXTENDING.md)** has the IR contract, a complete worked example
+(BBCode in ~40 lines), and a checklist.
+
+## Lineage
+
+Standing on the shoulders of the tools and documentation that kept WordStar
+readable: Yohanes Nugroho's WS-CON, Michael Petrie's English port, the `wsconvert`
+project, Robert J. Sawyer's WordStar archive, and the WordStar format
+documentation community. Behaviors were studied and reimplemented; no code was
+copied. The development corpus is personal and is not distributed — tests use
+synthetic fixtures that encode the same behaviors.
+
+## Credits
+
+Written by Jon Michaels — whose 1987–1992 WordStar files, and the need to read
+them again, are the reason this exists — with Athena (Claude, Anthropic) as
+co-author: the byte archaeology, the wrap test, and the implementation grew out
+of a joint effort to recover those disks. Every commit carries the co-author
+trailer.
+
+## License
+
+MIT © Jon Michaels

ctrl_kd-1.1.2/pyproject.toml

@@ -0,0 +1,30 @@
+[build-system]
+requires = ["setuptools>=68"]
+build-backend = "setuptools.build_meta"
+
+[project]
+name = "ctrl-kd"
+version = "1.1.2"
+description = "Convert WordStar 4-7 documents and print-to-disk files to text, Markdown, HTML, RTF, or PDF. ^KD: save and done."
+readme = "README.md"
+license = {text = "MIT"}
+authors = [{name = "Jon Michaels"}]
+requires-python = ">=3.9"
+keywords = ["wordstar", "converter", "retrocomputing", "archive", "dos"]
+classifiers = [
+    "Development Status :: 5 - Production/Stable",
+    "Environment :: Console",
+    "License :: OSI Approved :: MIT License",
+    "Programming Language :: Python :: 3",
+    "Topic :: Text Processing :: Filters",
+    "Topic :: System :: Archiving",
+]
+
+[project.urls]
+Homepage = "https://github.com/jonmichaels/ctrl-kd"
+
+[project.scripts]
+ctrl-kd = "ctrlkd.cli:main"
+
+[tool.setuptools.packages.find]
+where = ["src"]

ctrl_kd-1.1.2/setup.cfg

@@ -0,0 +1,4 @@
+[egg_info]
+tag_build = 
+tag_date = 0
+

ctrl_kd-1.1.2/src/ctrl_kd.egg-info/PKG-INFO

@@ -0,0 +1,120 @@
+Metadata-Version: 2.4
+Name: ctrl-kd
+Version: 1.1.2
+Summary: Convert WordStar 4-7 documents and print-to-disk files to text, Markdown, HTML, RTF, or PDF. ^KD: save and done.
+Author: Jon Michaels
+License: MIT
+Project-URL: Homepage, https://github.com/jonmichaels/ctrl-kd
+Keywords: wordstar,converter,retrocomputing,archive,dos
+Classifier: Development Status :: 5 - Production/Stable
+Classifier: Environment :: Console
+Classifier: License :: OSI Approved :: MIT License
+Classifier: Programming Language :: Python :: 3
+Classifier: Topic :: Text Processing :: Filters
+Classifier: Topic :: System :: Archiving
+Requires-Python: >=3.9
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Dynamic: license-file
+
+# ctrl-kd
+
+Convert WordStar-era files to modern formats. **^KD: save and done.**
+
+`ctrl-kd` reads WordStar 4 documents, WordStar 5–7 documents, and WordStar
+**print-to-disk files** (the printer byte stream, captured to a file — a distinct
+format most converters mangle), and writes plain text, Markdown, HTML, RTF, or PDF (typewriter-set on the
+built-in Courier fonts — no dependencies, the page as it would have printed).
+
+```console
+$ ctrl-kd ESSAY.WS                      # -> ESSAY.md
+$ ctrl-kd ESSAY.WS -t html -t rtf       # multiple formats
+$ ctrl-kd ESSAY.WS -t pdf --mode printed # a facsimile of the 1990 printout
+$ ctrl-kd --mode printed LETTER.WS      # line-for-line, as it printed in 1990
+$ ctrl-kd --diagnose MYSTERY.FIL        # what IS this file?
+```
+
+## Why another converter?
+
+Existing tools each lose something. Fed a WordStar 4 file, converters written for
+WS7 delete the last letter of every word (WS4 set bit 7 on it). Most delete soft
+returns outright — `Jon Michaels` + `March 6, 1992` becomes
+`Jon MichaelsMarch 6, 1992` — which also destroys every poem, because poem lines
+end in soft returns too. And print-to-disk files aren't WordStar documents at all,
+so feeding them to a WordStar converter produces stray superscripts and garbage.
+
+`ctrl-kd` was built by converting a real 1987–1992 corpus (high-school and college
+papers, poems, stories — WordStar 4 on DOS, dot-matrix printer) and verifying
+against surviving period printouts of the same documents. Its rules are empirical:
+
+* **Detection by content, never by extension.** WS4 vs WS5+ vs print stream vs
+  plain text vs binary, with the evidence shown in `--diagnose`.
+* **The wrap test.** WordStar wrapped only when the next word didn't fit. So a
+  soft return where the next word *would* have fit (strictly — WordStar wrapped
+  even on an exact-margin fit) is a deliberate break: a poem line, a heading.
+  Everything else is word wrap and joins with a space. The margin is estimated
+  from the 90th percentile of soft-wrapped line lengths (floor 65, the default).
+* **Break runs.** Soft/hard return runs containing a hard return and a blank line
+  are paragraph breaks; a lone hard return is the author's deliberate line break.
+  Double-spaced documents (blank soft lines between every line) collapse
+  automatically.
+* **Ruler lines mean columns.** A `.rr----!----` dot line defines tab stops; the
+  document's alignment is space-built and only survives fixed-width. Such
+  documents render `printed` in every mode.
+* **Print streams render verbatim** — they ARE the printed page — with printer
+  style codes decoded (superscript/underline/italic/bold pairs; table in
+  `core.PRINT_CODES`, derived from a late-80s dot-matrix driver and overridable).
+* **WS5+ symmetric blocks** (`0x1D`: real footnotes/endnotes, headings, page
+  breaks — machinery added in WS5) are parsed with their nested structure,
+  verified against the 86 WordStar 7 documents in Robert J. Sawyer's public
+  WordStar archive: footnotes extract with in-text references (`[^n]` in
+  Markdown), paragraph styles become headings, and 82/86 convert with zero
+  mojibake. More WS5–7 corpora still welcome.
+
+## Modes
+
+* `--mode modern` (default): reflowed paragraphs, semantic markup, deliberate
+  line breaks kept.
+* `--mode printed`: every line as laid out, fixed-width, `.pa`/form-feed page
+  breaks honored — how it came off the printer.
+
+## Install
+
+Straight from GitHub (not yet on PyPI):
+
+```console
+$ pipx install git+https://github.com/jonmichaels/ctrl-kd
+```
+
+or `pip install git+https://github.com/jonmichaels/ctrl-kd` into an environment
+of your choice. Python ≥ 3.9, no dependencies.
+Library API: `ctrlkd.convert(data, to='html')`.
+
+## Adding an output format
+
+An output format is one function over the parsed document — register it with the
+`@ctrlkd.emitter` decorator, or ship it as a pip-installable plugin via the
+`ctrlkd.emitters` entry-point group and it appears in the CLI automatically.
+**[EXTENDING.md](EXTENDING.md)** has the IR contract, a complete worked example
+(BBCode in ~40 lines), and a checklist.
+
+## Lineage
+
+Standing on the shoulders of the tools and documentation that kept WordStar
+readable: Yohanes Nugroho's WS-CON, Michael Petrie's English port, the `wsconvert`
+project, Robert J. Sawyer's WordStar archive, and the WordStar format
+documentation community. Behaviors were studied and reimplemented; no code was
+copied. The development corpus is personal and is not distributed — tests use
+synthetic fixtures that encode the same behaviors.
+
+## Credits
+
+Written by Jon Michaels — whose 1987–1992 WordStar files, and the need to read
+them again, are the reason this exists — with Athena (Claude, Anthropic) as
+co-author: the byte archaeology, the wrap test, and the implementation grew out
+of a joint effort to recover those disks. Every commit carries the co-author
+trailer.
+
+## License
+
+MIT © Jon Michaels

ctrl_kd-1.1.2/src/ctrl_kd.egg-info/SOURCES.txt

@@ -0,0 +1,14 @@
+LICENSE
+README.md
+pyproject.toml
+src/ctrl_kd.egg-info/PKG-INFO
+src/ctrl_kd.egg-info/SOURCES.txt
+src/ctrl_kd.egg-info/dependency_links.txt
+src/ctrl_kd.egg-info/entry_points.txt
+src/ctrl_kd.egg-info/top_level.txt
+src/ctrlkd/__init__.py
+src/ctrlkd/cli.py
+src/ctrlkd/core.py
+src/ctrlkd/emit.py
+src/ctrlkd/pdf.py
+tests/test_ctrlkd.py

ctrl_kd-1.1.2/src/ctrl_kd.egg-info/dependency_links.txt

@@ -0,0 +1 @@
+

ctrl_kd-1.1.2/src/ctrl_kd.egg-info/entry_points.txt

@@ -0,0 +1,2 @@
+[console_scripts]
+ctrl-kd = ctrlkd.cli:main

ctrl_kd-1.1.2/src/ctrl_kd.egg-info/top_level.txt

@@ -0,0 +1 @@
+ctrlkd

ctrl_kd-1.1.2/src/ctrlkd/__init__.py

@@ -0,0 +1,12 @@
+"""ctrl-kd — convert WordStar-era files to modern formats. ^KD: save and done."""
+from .core import detect, parse, parse_ws, parse_printstream, Document, Block, Line, Span
+from .emit import (emit_text, emit_markdown, emit_html, emit_rtf,
+                   emitter, get_emitter, formats, load_plugins)
+from .pdf import emit_pdf                        # registers the 'pdf' format
+
+__version__ = '1.1.2'
+
+def convert(data: bytes, to: str = 'markdown', mode: str = 'modern',
+            encoding: str = 'cp437', **options) -> str:
+    """One-call API: bytes in, converted string out."""
+    return get_emitter(to)['fn'](parse(data, encoding=encoding), mode, **options)

ctrl_kd-1.1.2/src/ctrlkd/cli.py

@@ -0,0 +1,90 @@
+"""ctrl-kd command line: convert WordStar-era files to modern formats.
+
+    ctrl-kd PAPER.WS                       # -> PAPER.md, modern reflow
+    ctrl-kd PAPER.WS -t html -o out.html
+    ctrl-kd --mode printed LETTER          # as it came off the printer
+    ctrl-kd --diagnose MYSTERY.FIL         # what IS this file?
+    ctrl-kd -t text -t html -d out/ *.WS   # batch, multiple formats
+"""
+import argparse, json, os, sys
+from . import core, emit
+
+def diagnose(path, data):
+    det = core.detect(data)
+    info = {'file': path, **det}
+    if det['variant'] in ('ws4', 'ws5+'):
+        doc = core.parse_ws(data)
+        info.update({k: doc.meta[k] for k in
+                     ('margin_estimate', 'dot_commands', 'unknown_codes', 'columnar')})
+        info['paragraphs'] = sum(1 for b in doc.blocks if b.kind == 'para')
+        info['footnotes'] = len(doc.footnotes)
+    return info
+
+def main(argv=None):
+    emit.load_plugins()          # third-party emitters (ctrlkd.emitters entry points)
+    ap = argparse.ArgumentParser(
+        prog='ctrl-kd',
+        description='Convert WordStar 4-7 documents and print-to-disk files to '
+                    'text, Markdown, HTML, or RTF (extensible: see EXTENDING.md). '
+                    '^KD: save and done.')
+    ap.add_argument('files', nargs='+', help='input file(s)')
+    ap.add_argument('-t', '--to', action='append', choices=emit.formats(),
+                    help='output format (repeatable; default: markdown)')
+    ap.add_argument('-o', '--output', help='output file (single input only)')
+    ap.add_argument('-d', '--outdir', help='output directory for batch conversion')
+    ap.add_argument('--mode', choices=('modern', 'printed'), default='modern',
+                    help='modern: reflowed paragraphs. printed: line-for-line, '
+                         'fixed-width, as it printed in 1990 (default: modern; '
+                         'print streams and ruler-line documents always render printed)')
+    ap.add_argument('--variant', choices=('ws4', 'ws5+', 'printstream', 'text'),
+                    help='override detection')
+    ap.add_argument('--encoding', default='cp437',
+                    help='byte encoding of the source (default: cp437)')
+    ap.add_argument('--diagnose', action='store_true',
+                    help='report what the file is (variant, margin, dot commands, '
+                         'unknown codes) as JSON; no conversion')
+    a = ap.parse_args(argv)
+    formats = a.to or ['markdown']
+    if a.output and (len(a.files) > 1 or len(formats) > 1):
+        ap.error('-o works with a single input and a single format; use -d for batch')
+
+    status = 0
+    for path in a.files:
+        try:
+            data = open(path, 'rb').read()
+        except OSError as e:
+            print(f'ctrl-kd: {e}', file=sys.stderr)
+            status = 1
+            continue
+        if a.diagnose:
+            print(json.dumps(diagnose(path, data), indent=2))
+            continue
+        try:
+            doc = core.parse(data, encoding=a.encoding, variant=a.variant)
+        except ValueError as e:
+            print(f'ctrl-kd: {path}: {e} (use --diagnose to inspect, '
+                  f'--variant to force)', file=sys.stderr)
+            status = 1
+            continue
+        base = os.path.splitext(os.path.basename(path))[0]
+        for fmt in formats:
+            reg = emit.get_emitter(fmt)
+            out = reg['fn'](doc, a.mode, title=base)
+            if a.output:
+                dest = a.output
+            else:
+                dest = os.path.join(a.outdir or os.path.dirname(path) or '.',
+                                    base + reg['ext'])
+                if a.outdir:
+                    os.makedirs(a.outdir, exist_ok=True)
+            if isinstance(out, bytes):           # binary formats (e.g. pdf)
+                with open(dest, 'wb') as f:
+                    f.write(out)
+            else:
+                with open(dest, 'w', encoding='utf-8', newline='\n') as f:
+                    f.write(out)
+            print(f'{path} -> {dest}')
+    return status
+
+if __name__ == '__main__':
+    sys.exit(main())