ctrl-kd 1.2.0__tar.gz → 1.3.0__tar.gz

{ctrl_kd-1.2.0/src/ctrl_kd.egg-info → ctrl_kd-1.3.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: ctrl-kd
-Version: 1.2.0
+Version: 1.3.0
 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
@@ -76,9 +76,11 @@ against surviving period printouts of the same documents. Its rules are empirica
   WordStar never printed them; `--diagnose` still reports that they exist.
   Paragraph styles become headings, and 82/86 convert with zero mojibake.
   More WS5–7 corpora still welcome.
-* **Page geometry** from the file's own `.pl`/`.po`/`.mt`/`.mb`, so `--mode
-  printed` uses the page the author set up rather than an assumed one — and
-  `--diagnose` says whether the size came from the file or from the default.
+* **Page geometry** from the file's own `.pl`/`.po`/`.mt`/`.mb`/`.hm`/`.fm`/
+  `.lh`/`.ls`, so `--mode printed` paginates with WordStar's own vertical model
+  — `.pl − .mt − .mb` at the `.lh` line height, 55 text lines for WordStar's
+  defaults, not a guessed 1-inch margin — and `--diagnose` says whether each
+  figure came from the file or from the default.
   In `printed` mode footnotes are laid out the way WordStar laid them out: at
   the foot of the page that references them, behind a twenty-dash separator,
   split across pages with `...Continued...` when they do not fit.

{ctrl_kd-1.2.0 → ctrl_kd-1.3.0}/README.md

@@ -57,9 +57,11 @@ against surviving period printouts of the same documents. Its rules are empirica
   WordStar never printed them; `--diagnose` still reports that they exist.
   Paragraph styles become headings, and 82/86 convert with zero mojibake.
   More WS5–7 corpora still welcome.
-* **Page geometry** from the file's own `.pl`/`.po`/`.mt`/`.mb`, so `--mode
-  printed` uses the page the author set up rather than an assumed one — and
-  `--diagnose` says whether the size came from the file or from the default.
+* **Page geometry** from the file's own `.pl`/`.po`/`.mt`/`.mb`/`.hm`/`.fm`/
+  `.lh`/`.ls`, so `--mode printed` paginates with WordStar's own vertical model
+  — `.pl − .mt − .mb` at the `.lh` line height, 55 text lines for WordStar's
+  defaults, not a guessed 1-inch margin — and `--diagnose` says whether each
+  figure came from the file or from the default.
   In `printed` mode footnotes are laid out the way WordStar laid them out: at
   the foot of the page that references them, behind a twenty-dash separator,
   split across pages with `...Continued...` when they do not fit.

{ctrl_kd-1.2.0 → ctrl_kd-1.3.0}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "ctrl-kd"
-version = "1.2.0"
+version = "1.3.0"
 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"}

{ctrl_kd-1.2.0 → ctrl_kd-1.3.0/src/ctrl_kd.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: ctrl-kd
-Version: 1.2.0
+Version: 1.3.0
 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
@@ -76,9 +76,11 @@ against surviving period printouts of the same documents. Its rules are empirica
   WordStar never printed them; `--diagnose` still reports that they exist.
   Paragraph styles become headings, and 82/86 convert with zero mojibake.
   More WS5–7 corpora still welcome.
-* **Page geometry** from the file's own `.pl`/`.po`/`.mt`/`.mb`, so `--mode
-  printed` uses the page the author set up rather than an assumed one — and
-  `--diagnose` says whether the size came from the file or from the default.
+* **Page geometry** from the file's own `.pl`/`.po`/`.mt`/`.mb`/`.hm`/`.fm`/
+  `.lh`/`.ls`, so `--mode printed` paginates with WordStar's own vertical model
+  — `.pl − .mt − .mb` at the `.lh` line height, 55 text lines for WordStar's
+  defaults, not a guessed 1-inch margin — and `--diagnose` says whether each
+  figure came from the file or from the default.
   In `printed` mode footnotes are laid out the way WordStar laid them out: at
   the foot of the page that references them, behind a twenty-dash separator,
   split across pages with `...Continued...` when they do not fit.

{ctrl_kd-1.2.0 → ctrl_kd-1.3.0}/src/ctrlkd/__init__.py

@@ -5,4 +5,4 @@ from .emit import (emit_text, emit_markdown, emit_html, emit_rtf,
 from .pdf import emit_pdf                        # registers the 'pdf' format
 from .convert import convert, select_notes, DEFAULT_NOTE_KINDS, ALL_NOTE_KINDS
 
-__version__ = '1.2.0'
+__version__ = '1.3.0'

{ctrl_kd-1.2.0 → ctrl_kd-1.3.0}/src/ctrlkd/core.py

@@ -16,6 +16,7 @@ WordStar background this code encodes:
   sent to the printer, captured to a file. They ARE the printed page.
 """
 from __future__ import annotations
+import math
 import re
 from dataclasses import dataclass, field
 
@@ -239,7 +240,9 @@ _DOT_CMD_RE = re.compile(rb'^\.([A-Za-z]{1,3})\s*(.*)$')
 _DOT_NUM_RE = re.compile(rb'^\s*([0-9]*\.?[0-9]+)\s*("|[A-Za-z]{1,2})?')
 
 _PAGE_DOT_KEYS = {b'PL': 'pl_lines', b'MT': 'mt_lines',
-                  b'MB': 'mb_lines', b'PO': 'po_cols'}
+                  b'MB': 'mb_lines', b'PO': 'po_cols',
+                  b'HM': 'hm_lines', b'FM': 'fm_lines',
+                  b'LH': 'lh_48', b'LS': 'ls'}
 
 # Named page sizes at 6 LPI (WordStar 7.0 file format spec: ".PL ... assuming
 # 6 lines per inch. An eleven inch page contains 66 lines."): 66 lines/11in
@@ -260,6 +263,10 @@ DEFAULT_MB_LINES = 8.0     # spec: ".MB ... The default value is 8 lines."
 DEFAULT_PO_COLS = 0.0      # no default is stated in the spec for .po; 0 (flush
                            # with the paper edge) is the least presumptuous
                            # reading rather than a remembered/guessed figure.
+DEFAULT_HM_LINES = 2.0     # spec: ".HM ... Default is 2." (header sits INSIDE .mt)
+DEFAULT_FM_LINES = 2.0     # spec: ".FM ... Default is 2." (footer sits INSIDE .mb)
+DEFAULT_LH_48 = 8.0        # spec: ".LH ... The default is 8/48 or 6 lines per inch."
+DEFAULT_LS = 1.0           # single spacing (WS7 manual, "Line Spacing")
 
 def _dot_arg_inches(value: float, unit: bytes | None):
     """Convert a dot-command argument's optional unit suffix to inches.
@@ -287,6 +294,57 @@ def _resolve_cols_arg(value: float, unit: bytes | None) -> float:
     inches = _dot_arg_inches(value, unit)
     return value if inches is None else inches * 10.0
 
+def _resolve_lh_arg(value: float, unit: bytes | None):
+    """.lh argument -> line height in 1/48in units. Unit-less IS 48ths (WS7
+    manual: "You can also type the dot command in 48ths of an inch. For
+    example, .lh 8 is 8/48 inch, or the standard 6 lines per inch"); an
+    explicit unit suffix converts. `.lh a` (auto-leading) never reaches here
+    -- the numeric matcher won't match it, so it stays default + verbatim.
+    A non-positive height is meaningless: rejected (None), default stands."""
+    inches = _dot_arg_inches(value, unit)
+    resolved = value if inches is None else inches * 48.0
+    return resolved if resolved > 0 else None
+
+def _resolve_ls_arg(value: float, unit: bytes | None):
+    """.ls argument -> line spacing. "A line spacing of between 1 and 9"
+    (WS7 file format spec); anything else is junk, rejected (None). Any unit
+    suffix is likewise junk -- spacing is a count, not a measure."""
+    if unit is not None or not 1 <= value <= 9:
+        return None
+    return value
+
+_PAGE_DOT_RESOLVERS = {'po_cols': _resolve_cols_arg, 'lh_48': _resolve_lh_arg,
+                       'ls': _resolve_ls_arg}     # everything else: lines at 6 LPI
+
+def _text_lines_per_page(pl_lines: float, mt_lines: float, mb_lines: float,
+                         lh_48: float) -> int:
+    """Printed text lines per page -- WordStar's own vertical model (WS7
+    manual, "Page Layout"): "The top and bottom margins define the space
+    between the text and the top and bottom of the paper. On an 8.5 x 11-inch
+    page, if the top margin is .33 inches and the bottom margin is 1.33
+    inches, the space left for text is 9.33 inches." Lines available is that
+    text height divided by the line height (.lh, 1/48in units): "Changing the
+    line height affects the number of lines that can be printed on a page."
+    WordStar's own defaults (.pl 66 .mt 3 .mb 8 .lh 8) give 55.
+
+    Deliberately NOT in the formula:
+    - .hm/.fm -- the header prints WITHIN .mt and the footer WITHIN .mb
+      (".MT ... The header is printed within this margin"; ".MB ... The
+      footer or page number is printed within this margin"), so they position
+      header/footer inside space already subtracted, never reserve more.
+    - .ls -- line-spacing blanks are literal lines in the file ("when you use
+      line spacing, the blank lines become part of the file", WS7 manual,
+      "Line Spacing"), so the body text already carries them; dividing
+      capacity by .ls would double-count.
+
+    Unit-less .mt/.mb are lines at the fixed 6 LPI baseline (the module-note
+    assumption); .lh at parse time is resolved once per document (first
+    occurrence wins), not tracked per-line."""
+    usable = pl_lines - mt_lines - mb_lines            # lines at 6 LPI
+    if not math.isfinite(usable) or not math.isfinite(lh_48) or lh_48 <= 0:
+        return 1
+    return max(1, int(usable * 8.0 / lh_48))
+
 def _resolve_page_size(pl_lines: float):
     """pl_lines -> (height_in, size_name). Snaps to a named size when close;
     otherwise reports the raw geometry under 'Custom' rather than forcing a
@@ -340,8 +398,10 @@ def _parse_page_dot(cmd: bytes, page: dict, meta_extra: dict):
             return
         value = float(num.group(1))
         unit = num.group(2)
-        resolver = _resolve_cols_arg if name == b'PO' else _resolve_lines_arg
-        page[key] = resolver(value, unit)
+        resolver = _PAGE_DOT_RESOLVERS.get(key, _resolve_lines_arg)
+        resolved = resolver(value, unit)
+        if resolved is not None:                      # junk argument: default stands
+            page[key] = resolved
     elif name in (b'PT', b'PSA', b'PSB'):
         # WordTsar's own invented dot commands (its source calls them "not a
         # Wordstar command"). A real WordStar file never contains these --
@@ -676,6 +736,10 @@ def parse_ws(data: bytes, encoding: str = 'cp437') -> Document:
     mt_lines = page.get('mt_lines')
     mb_lines = page.get('mb_lines')
     po_cols = page.get('po_cols')
+    hm_lines = page.get('hm_lines')
+    fm_lines = page.get('fm_lines')
+    lh_48 = page.get('lh_48')
+    ls = page.get('ls')
     # Exposed per the IR contract: a consumer must be able to distinguish
     # "Legal (from file)" from "Letter (default)" -- provenance lives
     # alongside every resolved figure, not just the page size.
@@ -690,7 +754,22 @@ def parse_ws(data: bytes, encoding: str = 'cp437') -> Document:
         'mb_source': 'file' if mb_lines is not None else 'default',
         'po_cols': po_cols if po_cols is not None else DEFAULT_PO_COLS,
         'po_source': 'file' if po_cols is not None else 'default',
+        'hm_lines': hm_lines if hm_lines is not None else DEFAULT_HM_LINES,
+        'hm_source': 'file' if hm_lines is not None else 'default',
+        'fm_lines': fm_lines if fm_lines is not None else DEFAULT_FM_LINES,
+        'fm_source': 'file' if fm_lines is not None else 'default',
+        'lh_48': lh_48 if lh_48 is not None else DEFAULT_LH_48,
+        'lh_source': 'file' if lh_48 is not None else 'default',
+        'ls': ls if ls is not None else DEFAULT_LS,
+        'ls_source': 'file' if ls is not None else 'default',
     }
+    # The one derived figure consumers actually need: printed text lines per
+    # page, from WordStar's own vertical model (see _text_lines_per_page for
+    # the formula and the deliberate exclusions). Defaults -> 55, NOT the 60
+    # a naive 1in-margin Letter computation gives.
+    doc.meta['page']['text_lines'] = _text_lines_per_page(
+        doc.meta['page']['pl_lines'], doc.meta['page']['mt_lines'],
+        doc.meta['page']['mb_lines'], doc.meta['page']['lh_48'])
     if meta_extra:
         doc.meta.update(meta_extra)
     return doc

{ctrl_kd-1.2.0 → ctrl_kd-1.3.0}/src/ctrlkd/pdf.py

@@ -19,12 +19,14 @@ PAGE_W, PAGE_H = 612, 792            # US Letter, points
 MARGIN = 72                          # 1 inch
 SIZE, LEAD = 12, 12                  # 10 CPI pica x 6 LPI — the dot-matrix standard;
                                      # a 65-col WordStar line is exactly 6.5in
-TOP_MODERN, TOP_PRINTED = 72, 36     # print streams carry their own top-margin blanks
+TOP_MODERN, TOP_PRINTED = 72, 36     # printed: default when a stream has no geometry
+                                     # meta (its margin blanks travel in-band); WS docs
+                                     # get an .mt-derived top from _printed_top()
 LINES_MODERN = (PAGE_H - 2 * 72) // LEAD                 # 54
-LINES_PRINTED = (PAGE_H - 2 * 36) // LEAD                # 60 -- the US Letter default;
-                                                          # printed mode's real per-document
-                                                          # figure comes from _printed_cap(),
-                                                          # which honours .pl-derived geometry
+# Printed capacity is per-document: _printed_cap() -- WordStar's own model
+# (.pl - .mt - .mb at the .lh line height; 55 for WordStar's defaults). The
+# old hardcoded (PAGE_H - 2*36)//LEAD = 60 was a naive Letter computation
+# that matched no WordStar the manual describes.
 MAX_COLS = int((PAGE_W - 2 * MARGIN) / (SIZE * 0.6))     # 65 — WordStar's own margin
 
 # Period-authentic footnote layout (Printed mode only -- WordStar Professional
@@ -55,9 +57,45 @@ def _resolved_page_height(doc, printed):
 def _printed_cap(doc):
     """Lines of vertical room on a printed page for THIS document -- the
     cap used both for plain pagination and as the footnote layout's page
-    budget (see _paginate_printed_notes)."""
+    budget (see _paginate_printed_notes).
+
+    WS documents carry doc.meta['page'] and get WordStar's own vertical
+    model (core._text_lines_per_page: .pl - .mt - .mb at the .lh line
+    height -- 55 for WordStar's defaults, NOT the 60 a naive 1in-margin
+    computation gives). Print streams have no 'page' meta and ARE the
+    printed page -- their margin blanks travel in-band -- so their budget
+    is the FULL page height in lines (66 on Letter): anything smaller
+    would split a physical page that the printer produced whole."""
+    page = doc.meta.get('page')
+    if page is not None:
+        return max(FOOTNOTE_FLOOR + 1, page.get('text_lines', 55))
     page_h = _resolved_page_height(doc, True)
-    return max(FOOTNOTE_FLOOR + 1, (page_h - 2 * TOP_PRINTED) // LEAD)
+    return max(FOOTNOTE_FLOOR + 1, int(page_h // LEAD))
+
+def _printed_top(doc):
+    """Top-of-text offset in points for printed mode. WS documents start
+    where .mt says (lines at 6 LPI -> 12pt each; the default .mt 3 is the
+    36pt this emitter always used). Print streams keep the fixed 36pt --
+    their own top-margin blanks are in the data (minus the machine-margin
+    strip in _doc_to_pagelines). Clamped inside the page so garbage .mt
+    from a misdetected binary degrades to an ugly page, never an absurd
+    coordinate space."""
+    page = doc.meta.get('page')
+    if page is None:
+        return TOP_PRINTED
+    page_h = _resolved_page_height(doc, True)
+    return max(0, min(round(page.get('mt_lines', 3.0) * 12), page_h - LEAD))
+
+def _printed_lead(doc):
+    """Baseline-to-baseline distance in points for printed mode: .lh is
+    1/48in units, a point is 1/72in -> lh * 1.5. Default .lh 8 IS the 12pt
+    lead this emitter always used. Print streams (no 'page' meta) keep the
+    fixed LEAD."""
+    page = doc.meta.get('page')
+    if page is None:
+        return LEAD
+    lh = page.get('lh_48', 8.0)
+    return lh * 1.5 if lh > 0 else LEAD
 
 FONTS = {(False, False): 'F1', (True, False): 'F2',
          (False, True): 'F3', (True, True): 'F4'}
@@ -416,7 +454,7 @@ def _coalesce(line):
             out.append([text, styles])
     return out
 
-def _page_stream(pagelines, top, page_h=PAGE_H):
+def _page_stream(pagelines, top, page_h=PAGE_H, lead=LEAD):
     ops = []
     y = page_h - top - SIZE
     for line in pagelines:
@@ -436,7 +474,7 @@ def _page_stream(pagelines, top, page_h=PAGE_H):
             if 'strike' in styles and text.strip():
                 ops.append(b'0.6 w %.1f %.1f m %.1f %.1f l S' % (x, y + 3, x + w, y + 3))
             x += w
-        y -= LEAD
+        y -= lead
     return b'\n'.join(ops)
 
 @emitter('pdf')
@@ -445,7 +483,9 @@ def emit_pdf(doc, mode='modern', **options):
     stream per page, xref. Returns bytes — PDF is a binary format."""
     printed = mode == 'printed' or _printed(doc)
     pages = _doc_to_pagelines(doc, printed)
-    top = TOP_PRINTED if printed else TOP_MODERN
+    top = _printed_top(doc) if printed else TOP_MODERN    # .mt-derived for WS docs;
+                                                           # default .mt 3 IS the old 36pt
+    lead = _printed_lead(doc) if printed else LEAD        # .lh-derived; .lh 8 IS 12pt
     page_h = _resolved_page_height(doc, printed)          # file geometry wins in
                                                            # printed mode (Task: .pl);
                                                            # modern stays fixed Letter
@@ -476,7 +516,7 @@ def emit_pdf(doc, mode='modern', **options):
                      b'<< /Type /Page /Parent 2 0 R /MediaBox [0 0 %d %d] '
                      b'/Resources << /Font << %s >> >> /Contents %d 0 R >>'
                      % (PAGE_W, page_h, font_dict, cnum)))
-        stream = _page_stream(pl, top, page_h)
+        stream = _page_stream(pl, top, page_h, lead)
         objs.append((cnum, b'<< /Length %d >>\nstream\n%s\nendstream'
                      % (len(stream), stream)))
 

{ctrl_kd-1.2.0 → ctrl_kd-1.3.0}/tests/test_ctrlkd.py

@@ -778,6 +778,115 @@ def test_page_geometry_malformed_pl_does_not_crash():
     assert doc.meta['page']['pl_lines'] == 66.0
     assert doc.meta['page']['size_source'] == 'default'
 
+# ------------------------------------------- the vertical model (text_lines)
+
+def test_page_geometry_defaults_give_wordstar_55_not_60():
+    # THE fix this section exists for: WordStar's own defaults (.pl 66
+    # .mt 3 .mb 8 .lh 8) put 55 text lines on a page -- the manual's model,
+    # (pl - mt - mb) at 6 LPI -- not the 60 a naive 1in-margin Letter
+    # computation produced for every document before 1.3.0.
+    from ctrlkd.pdf import _printed_cap
+    doc = core.parse_ws(b'Body text.' + HARD)
+    assert doc.meta['page']['text_lines'] == 55
+    assert _printed_cap(doc) == 55
+
+def test_page_geometry_mt_mb_change_capacity():
+    doc = core.parse_ws(b'.MT 6' + HARD + b'.MB 6' + HARD + b'x' + HARD)
+    assert doc.meta['page']['text_lines'] == 54          # 66 - 6 - 6
+
+def test_page_geometry_lh_parsed_and_scales_capacity():
+    # .lh is 1/48in units: .lh 16 doubles the line height, halving capacity
+    # (the manual: "Changing the line height affects the number of lines
+    # that can be printed on a page"). floor(55 * 8 / 16) = 27.
+    doc = core.parse_ws(b'.LH 16' + HARD + b'x' + HARD)
+    page = doc.meta['page']
+    assert page['lh_48'] == 16.0 and page['lh_source'] == 'file'
+    assert page['text_lines'] == 27
+
+def test_page_geometry_lh_unit_suffix_converts():
+    # .lh 12p = 12/72in = 8/48in -> the standard height, stated in points
+    doc = core.parse_ws(b'.LH 12P' + HARD + b'x' + HARD)
+    assert doc.meta['page']['lh_48'] == 8.0
+    assert doc.meta['page']['text_lines'] == 55
+
+def test_page_geometry_lh_zero_and_auto_rejected():
+    # .lh 0 is meaningless and .lh a is auto-leading -- both leave the
+    # default standing (and stay preserved verbatim in dot_commands)
+    for arg in (b'.LH 0', b'.LH A'):
+        doc = core.parse_ws(arg + HARD + b'x' + HARD)
+        assert doc.meta['page']['lh_48'] == 8.0
+        assert doc.meta['page']['lh_source'] == 'default'
+        assert doc.meta['page']['text_lines'] == 55
+
+def test_page_geometry_ls_recorded_but_never_divides_capacity():
+    # the trap the manual defuses: line-spacing blanks are LITERAL lines in
+    # the file ("when you use line spacing, the blank lines become part of
+    # the file" -- WS7 manual, "Line Spacing"), so the body already carries
+    # them; dividing capacity by .ls would double-count.
+    doc = core.parse_ws(b'.LS 2' + HARD + b'x' + HARD)
+    page = doc.meta['page']
+    assert page['ls'] == 2.0 and page['ls_source'] == 'file'
+    assert page['text_lines'] == 55                       # unchanged
+
+def test_page_geometry_ls_out_of_range_rejected():
+    # spec: "a line spacing of between 1 and 9"
+    for arg in (b'.LS 0', b'.LS 12'):
+        doc = core.parse_ws(arg + HARD + b'x' + HARD)
+        assert doc.meta['page']['ls'] == 1.0
+        assert doc.meta['page']['ls_source'] == 'default'
+
+def test_page_geometry_hm_fm_parsed_but_reserve_no_space():
+    # header/footer margins position header and footer INSIDE .mt/.mb
+    # (".MT ... The header is printed within this margin") -- parsed with
+    # provenance for --diagnose, never subtracted from capacity
+    doc = core.parse_ws(b'.HM 1' + HARD + b'.FM 3' + HARD + b'x' + HARD)
+    page = doc.meta['page']
+    assert page['hm_lines'] == 1.0 and page['hm_source'] == 'file'
+    assert page['fm_lines'] == 3.0 and page['fm_source'] == 'file'
+    assert page['text_lines'] == 55                       # unchanged
+
+def test_page_geometry_absurd_margins_clamp_not_crash():
+    # margins that eat the whole page (garbage in a misdetected binary)
+    # degrade to a 1-line model, never zero/negative/crash
+    doc = core.parse_ws(b'.PL 12' + HARD + b'.MT 40' + HARD + b'.MB 40' + HARD + b'x' + HARD)
+    assert doc.meta['page']['text_lines'] == 1
+
+def test_pdf_printed_top_offset_follows_mt():
+    # the default .mt 3 IS the 36pt top this emitter always used; a bigger
+    # .mt moves the text start down in real points (1 line = 12pt at 6 LPI)
+    from ctrlkd.pdf import _printed_top
+    assert _printed_top(core.parse_ws(b'x' + HARD)) == 36
+    assert _printed_top(core.parse_ws(b'.MT 6' + HARD + b'x' + HARD)) == 72
+
+def test_pdf_printed_lead_follows_lh():
+    # .lh 8 IS the 12pt lead; .lh 16 prints double-spaced at 24pt
+    from ctrlkd.pdf import _printed_lead
+    assert _printed_lead(core.parse_ws(b'x' + HARD)) == 12.0
+    assert _printed_lead(core.parse_ws(b'.LH 16' + HARD + b'x' + HARD)) == 24.0
+
+def test_pdf_output_bytes_carry_mt_top_and_lh_lead():
+    # end-to-end: the geometry must reach the CONTENT STREAM, not just the
+    # helpers -- .mt 6 starts text at 72pt from the top, .lh 16 spaces
+    # baselines 24pt apart. Read the Td y-coordinates back out of the bytes.
+    import re
+    from ctrlkd.pdf import emit_pdf
+    data = (b'.MT 6' + HARD + b'.LH 16' + HARD +
+            b'Line one.' + HARD + b'Line two.' + HARD + b'Line three.' + HARD)
+    pdf = emit_pdf(core.parse_ws(data), mode='printed')
+    ys = [float(m) for m in re.findall(rb'[\d.]+ ([\d.]+) Td', pdf)]
+    assert ys[0] == 792 - 72 - 12                  # top from .mt, not fixed 36
+    assert ys[0] - ys[1] == 24.0                   # lead from .lh, not fixed 12
+    assert ys[1] - ys[2] == 24.0
+
+def test_pdf_printstream_capacity_is_the_full_page():
+    # a print stream IS the printed page -- its margin blanks travel in-band,
+    # so the page budget is the FULL 66 lines of a Letter page: anything
+    # smaller could split a physical page the printer produced whole
+    from ctrlkd.pdf import _printed_cap, _printed_top
+    doc = core.parse_printstream(b'line one\r\nline two\r\n')
+    assert _printed_cap(doc) == 66
+    assert _printed_top(doc) == 36                        # fixed: not .mt-derived
+
 # ---------------------------------------------------------------- small parser additions
 
 def _ws7_tab(size_hmi, tab_type_byte, tenths=0):
@@ -874,6 +983,8 @@ def test_pdf_printed_footnote_splits_with_continuation_and_loses_nothing():
     from ctrlkd.pdf import _doc_to_pagelines, CONTINUATION_TEXT
     words = [f'word{i:03d}' for i in range(80)]
     data = (b'.PL 18' + HARD +                    # a small page: forces a split
+            b'.MT 3' + HARD + b'.MB 3' + HARD +   # stated so capacity is 18-3-3=12,
+                                                   # not left to the 3+8 defaults
             ws7_block(0x00) +
             b'First body line has the note' + ws7_note(0x03, ' '.join(words).encode(), number=0) +
             b' right here.' + HARD +
@@ -1041,7 +1152,9 @@ def test_pdf_printed_no_page_ever_exceeds_its_capacity():
     # once even one body line has been placed on a terminal page.
     from ctrlkd.pdf import _doc_to_pagelines, _printed_cap
     words = [f'word{i:03d}' for i in range(60)]
-    data = (b'.PL 12' + HARD + ws7_block(0x00) +
+    data = (b'.PL 12' + HARD +
+            b'.MT 3' + HARD + b'.MB 3' + HARD +   # capacity 12-3-3=6, margins stated
+            ws7_block(0x00) +
             b'Only line has a note' + ws7_note(0x03, ' '.join(words).encode(), number=0) +
             b' here.' + HARD)
     doc = core.parse_ws(data)