ctrl-kd 4.7.1__tar.gz → 4.8.1__tar.gz

{ctrl_kd-4.7.1/src/ctrl_kd.egg-info → ctrl_kd-4.8.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: ctrl-kd
-Version: 4.7.1
+Version: 4.8.1
 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

{ctrl_kd-4.7.1 → ctrl_kd-4.8.1/src/ctrl_kd.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: ctrl-kd
-Version: 4.7.1
+Version: 4.8.1
 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

{ctrl_kd-4.7.1 → ctrl_kd-4.8.1}/src/ctrl_kd.egg-info/SOURCES.txt

@@ -38,6 +38,7 @@ tests/test_form_feed_line_marks.py
 tests/test_glyph_aspect.py
 tests/test_graphic_cell_ops.py
 tests/test_graphic_cells.py
+tests/test_head_foot_placement_m16.py
 tests/test_head_foot_timing.py
 tests/test_html_print_stylesheet.py
 tests/test_justification.py
@@ -57,10 +58,13 @@ tests/test_merge_page_number_variable.py
 tests/test_modern_blank_after_tightened_line.py
 tests/test_modern_box_regions.py
 tests/test_modern_centring_clip.py
+tests/test_modern_columns.py
 tests/test_modern_line_spacing.py
 tests/test_modern_lint.py
 tests/test_modern_page_baseline.py
+tests/test_modern_page_number.py
 tests/test_modern_rtf_structure_rows.py
+tests/test_modern_sheet_height.py
 tests/test_modern_verse_defrow.py
 tests/test_note_rulings_20260824.py
 tests/test_page_parity.py
@@ -94,6 +98,8 @@ tests/test_screenplay_rendering.py
 tests/test_sentence_spacing_n9.py
 tests/test_structural_checks.py
 tests/test_style_leading.py
+tests/test_style_strikeout_runs_until_cleared.py
+tests/test_style_symbol_map_resolved_face.py
 tests/test_symbol_span_graphic_chars.py
 tests/test_trailing_pa_break.py
 tests/test_verse_quote_couplet.py

{ctrl_kd-4.7.1 → ctrl_kd-4.8.1}/src/ctrlkd/__init__.py

@@ -8,4 +8,4 @@ from .pdf import emit_pdf                        # registers the 'pdf' format
 from .convert import convert, select_notes, DEFAULT_NOTE_KINDS, ALL_NOTE_KINDS
 from .info import document_info
 
-__version__ = '4.7.1'
+__version__ = '4.8.1'

{ctrl_kd-4.7.1 → ctrl_kd-4.8.1}/src/ctrlkd/cli.py

@@ -130,15 +130,17 @@ def main(argv=None):
                     help='omit paragraph-style pass-through (HTML classes + '
                          'generated CSS, RTF stylesheet) from the output')
     ap.add_argument('--headers', choices=('on', 'off'), default='on',
-                    help='headers, footers, and page numbers in the paged '
-                         'surfaces (Printed/Native PDF and RTF). Default: on')
+                    help="the document's own running heads and feet (.he/.fo) "
+                         'in the paged surfaces -- Printed PDF, Modern PDF and '
+                         "both RTF modes. WordStar's own automatic page number "
+                         'is --page-numbers\' business alone. Default: on')
     ap.add_argument('--page-numbers', choices=('auto', 'on', 'off'),
                     default='auto',
                     help="WordStar's own AUTOMATIC page number -- the one "
                          '.pc positions, a separate mechanism from a # the '
-                         'author placed inside a real .he/.fo (that always '
-                         'prints, unaffected by this flag). Printed PDF '
-                         'and both RTF modes. auto (DEFAULT): the '
+                         'author placed inside a real .he/.fo (that goes '
+                         'with its head, under --headers, never this flag). '
+                         'Every paged surface. auto (DEFAULT): the '
                          'document\'s own dot '
                          'commands decide -- .pn/.pg turn it on, .op turns '
                          'it off, exactly like real WordStar; a document '
@@ -149,8 +151,9 @@ def main(argv=None):
                          'silent document. off: suppress it unconditionally. '
                          'A declared footer always pre-empts it, in every '
                          'mode (WSFORMAT.WS: "active only when the footers '
-                         'are not in use"). --headers off also suppresses '
-                         'it, per --headers\' own documented scope.')
+                         'are not in use") -- a property of the document, so '
+                         '--headers off does not lift it, and does not '
+                         'suppress the number either.')
     ap.add_argument('--line-numbers', choices=('on', 'off'), default='on',
                     help="the document's own .l# line-number gutter in the "
                          'paged surfaces (Printed/Native PDF and RTF); no '

{ctrl_kd-4.7.1 → ctrl_kd-4.8.1}/src/ctrlkd/core.py

@@ -1427,6 +1427,46 @@ class Document:
     # can `|` it into a run's own styles unconditionally.
     header_style_attrs: dict = field(default_factory=dict)
     footer_style_attrs: dict = field(default_factory=dict)
+    # M16 (2026-09-15, WS7 probes B0-B5 against sawyer/REF/BOOKLET.WS): the
+    # RIGHT MARGIN the head/foot line's own style declares, in print
+    # columns at 10 CPI (`right_margin_hmi / 180`), or absent when the
+    # style declares none. `{line: cols}`.
+    #
+    # A right- or centre-aligned running head aligns against THIS, measured
+    # from `.po` -- not against the document's `.rm`, not against the
+    # column grid, and not against the sheet. Measured, five probes printed
+    # by real WS7 under DOSBox-X: turning `.co 2` off (B2) and changing it
+    # to `.co 3` with a `.rm` of 2.50" instead of 4.50" (B3) both leave
+    # BOOKLET.WS's "Header Odd 1" at exactly the x the unmodified document
+    # prints it at, while moving `.po` from 0.2in to 1.2in (B4) moves it by
+    # exactly 1.00in. Its style's own `right_margin_hmi` is 18000 = 100
+    # columns = 10.00in, and `.po` 0.2in + 10.00in = 10.20in is where the
+    # last glyph of "Header Odd 1" ends, to 0.5pt.
+    #
+    # Every other document in the corpus with a right-aligned head declares
+    # a style right margin EQUAL to its own `.rm` (a "Header at Top of
+    # Page" style at 11700 = 6.50in beside a `.rm 65`), which is why the
+    # `.rm` reading was right everywhere else and wrong here.
+    header_style_rm: dict = field(default_factory=dict)
+    footer_style_rm: dict = field(default_factory=dict)
+    # M16: the LINE HEIGHT in force where this head/foot line was DEFINED,
+    # in 48ths of an inch -- the step between successive running-head
+    # lines, and 0 when the document asked for zero. `{line: lh_48}`.
+    #
+    # Measured (probe B1): removing BOOKLET.WS's own `.lh 0` -- which sits
+    # immediately before its `.f1`/`.h1`/`.h2` and is cancelled by a
+    # `.lh12` immediately after them -- makes real WS7 print its two header
+    # lines on two rows 1/6in apart instead of both on one row, with the
+    # columns left in place. Turning the columns off instead (B2) leaves
+    # them on one row. So the collapse is the line height's doing, not the
+    # column grid's, and it is the value in force at the COMMAND (the
+    # `.lh12` that follows governs the body and not these lines).
+    #
+    # The block is anchored on its LAST line either way (probe B5: a
+    # one-line head prints on the same row the second of two occupies), so
+    # a zero step stacks every line onto that row.
+    header_leads: dict = field(default_factory=dict)
+    footer_leads: dict = field(default_factory=dict)
     # Planning #250: `.H1E`/`.H1O`/`.F1E`/`.F1O` (WSFORMAT.TXT: ".HE can
     # optionally specify even or odd numbered page headers" / the matching
     # ".FO" sentence -- confirmed the ONLY two lines that do; `.H2`-`.H5`/
@@ -3273,7 +3313,18 @@ def _parse_format_dot(cmd: bytes, state: dict) -> None:
                 resolved = _resolve_lh_arg(value, m.group(2))
                 if resolved is not None:     # junk/non-positive: state stands
                     state['lead_48'] = resolved
+                    state['hf_lead_48'] = resolved
                     state['lh_auto'] = False
+                elif value == 0:
+                    # M16 (2026-09-15): `.lh 0` leaves `lead_48` alone --
+                    # a zero body leading is not something any renderer or
+                    # the pagination arithmetic can use, and no evidence
+                    # says WordStar applies it to the body -- but it IS
+                    # what a running head's own line step becomes, proved
+                    # by probes B0/B1/B2 against real WS7 (see
+                    # `Document.header_leads`). Recorded on its own key so
+                    # the head can read it and nothing else changes.
+                    state['hf_lead_48'] = 0.0
     elif name in (b'LM', b'RM', b'PM'):     # left / right / paragraph margin
         # Print columns at 10 CPI, matching `.po`; a unit suffix converts. The
         # archive writes both (`.rm 65` and `.rm 6.5"`).
@@ -3637,7 +3688,8 @@ def _hf_within(doc, cur, cur_line):
 
 def _parse_head_foot(cmd: bytes, doc, encoding: str, anchor=None, font_idx=None,
                      tab_mark=None, align=None, style_attrs=None,
-                     pctl_marks=None, within=None):
+                     pctl_marks=None, within=None, style_rm_cols=None,
+                     lead_48=None):
     """Record `.he`/`.h1`-`.h5` and `.fo`/`.f1`-`.f5` text on the Document.
 
     `.HE` and `.FO` are line 1; the numbered forms select their own line, so a
@@ -3766,6 +3818,18 @@ def _parse_head_foot(cmd: bytes, doc, encoding: str, anchor=None, font_idx=None,
     which_style_attrs = (doc.header_style_attrs if kind == 'H'
                          else doc.footer_style_attrs)
     which_style_attrs[line] = style_attrs or frozenset()
+    # M16: see `Document.header_style_rm`/`header_leads` for the probes.
+    which_style_rm = (doc.header_style_rm if kind == 'H'
+                      else doc.footer_style_rm)
+    if style_rm_cols is not None:
+        which_style_rm[line] = style_rm_cols
+    else:
+        which_style_rm.pop(line, None)
+    which_leads = doc.header_leads if kind == 'H' else doc.footer_leads
+    if lead_48 is not None:
+        which_leads[line] = lead_48
+    else:
+        which_leads.pop(line, None)
     if parity is not None:
         which_p = doc.headers_parity if kind == 'H' else doc.footers_parity
         which_fonts_p = (doc.header_fonts_parity if kind == 'H'
@@ -5162,6 +5226,26 @@ def _symmetric_blocks(data: bytes, encoding: str, raw_out=None):
             includes, driver[0], sorted(shift_runs), marks, header,
             pcl_programs)
 
+# The style attributes that KEEP RUNNING past the paragraph that turned them
+# on, until a later style's own attrs_OFF word clears them (ruling 2026-09-16,
+# WordStar-Feature-Decision-Register, "Style-library strikeout runs until a
+# style clears it").
+#
+# WSFORMAT.TXT states the inherit rule for the attribute words as a whole --
+# "if both corresponding bits are off, then the attribute is inherited from
+# the current state" -- but the only attribute the WS7 LaserJet captures can
+# actually DEMONSTRATE it for is strikeout: in every corpus document that
+# turns bold, underline or italic on in a style, the next style used sets
+# that same bit in its own attrs_off word, so an inheriting model and a
+# reset-per-paragraph model produce identical pages and the captures cannot
+# choose between them. Strikeout is the one bit no corpus style ever clears,
+# which is exactly why its run is visible (NOVEL.WS: 954 overstrike dash rows
+# across all 52 captured pages). So the rule is applied to strikeout only,
+# on measured evidence, and the wider question is left open rather than
+# guessed. Add a bit here only with a capture that shows it.
+STICKY_STYLE_ATTRS = ((0x01, 'strike'),)
+
+
 def _parse_style_library(raw: bytes, base: int, encoding: str = 'cp437'):
     """The paragraph style library at file-absolute offset `base`.
 
@@ -6146,6 +6230,7 @@ def parse_ws(data: bytes, encoding: str = 'cp437') -> Document:
             hf_align = None
             hf_style_font_idx = None
             hf_style_attrs = None
+            hf_style_rm = None
             if hf_style_w0 is not None and (hf_style_w0 >> 8) == 0x02:
                 _hf_entry = style_slots.get(hf_style_w0 & 0xFF)
                 if _hf_entry is not None:
@@ -6153,6 +6238,11 @@ def parse_ws(data: bytes, encoding: str = 'cp437') -> Document:
                     if _hf_entry.get('font') and any(_hf_entry['font']):
                         hf_style_font_idx = _style_font(_hf_entry['font'])
                     hf_style_attrs = _hf_entry.get('attrs')
+                    # M16: the same 180-HMI-per-column conversion
+                    # `emit.py` already applies to a BODY block's own
+                    # style right margin, read here for the head.
+                    if _hf_entry.get('right_margin_hmi'):
+                        hf_style_rm = _hf_entry['right_margin_hmi'] / 180.0
             _parse_head_foot(cmd if strip_hibit else raw.rstrip(), doc,
                              encoding,
                              anchor=len(doc.blocks) + (1 if cur.lines or
@@ -6175,7 +6265,16 @@ def parse_ws(data: bytes, encoding: str = 'cp437') -> Document:
                                       else hf_style_font_idx),
                              tab_mark=hf_tab_mark,
                              align=hf_align, style_attrs=hf_style_attrs,
-                             pctl_marks=hf_pctl_marks)
+                             pctl_marks=hf_pctl_marks,
+                             style_rm_cols=hf_style_rm,
+                             # M16: the `.lh` in force RIGHT HERE, which is
+                             # what the step between running-head lines
+                             # turns out to be -- `hf_lead_48` and not
+                             # `lead_48`, because a `.lh 0` leaves the
+                             # latter standing (it is not a body leading
+                             # any renderer could use) while still being
+                             # the head's own answer.
+                             lead_48=fmt.get('hf_lead_48'))
             # The index of the block this entry POINTS AT -- the one that follows it,
             # which is the block still open (if it has content) or the next to open.
             # "This heading is in the table of contents" refers forward, not back.
@@ -6422,8 +6521,34 @@ def parse_ws(data: bytes, encoding: str = 'cp437') -> Document:
                     # value forward, not reset to nothing -- style_fmt.clear()
                     # below must not lose it.
                     _prev_vmi = style_fmt.get('line_height_vmi')
+                    # Ruling 2026-09-16 ("Style-library strikeout runs until a
+                    # style clears it"): an attribute a style turns ON stays on
+                    # past that paragraph. WSFORMAT.TXT's own rule for the two
+                    # attribute words is "if both corresponding bits are off,
+                    # then the attribute is inherited from the current state" --
+                    # so only a later style that sets the bit in its attrs_OFF
+                    # word ends the run. Real WS7 agrees: NOVEL.WS's H3 sets
+                    # strikeout on the title page and nothing in that file's
+                    # library ever clears it, and the LaserJet capture prints
+                    # the overstrike dash row on all 52 pages.
+                    _prev_sticky = style_fmt.get('sticky_attrs', frozenset())
                     style_fmt.clear()
                     style_fmt['style_id'] = slot
+                    _sticky = set(_prev_sticky)
+                    if entry is not None:
+                        _on, _off = entry.get('attrs_on') or 0, entry.get('attrs_off') or 0
+                        for _bit, _tag in STICKY_STYLE_ATTRS:
+                            if _on & _bit:
+                                _sticky.add(_tag)
+                            elif _off & _bit:
+                                _sticky.discard(_tag)
+                    style_fmt['sticky_attrs'] = frozenset(_sticky)
+                    if _sticky:
+                        # Also for an UNRESOLVABLE handle (a 0x03xx editing-temp
+                        # pool entry, or a slot this file's library doesn't
+                        # carry): it declares neither word, so every attribute
+                        # inherits, including the running one.
+                        style_fmt['attrs'] = frozenset(_sticky)
                     if entry is not None:
                         style_fmt['style_name'] = entry['name']
                         style_fmt['heading'] = _style_heading_level(entry['name'])
@@ -6443,8 +6568,11 @@ def parse_ws(data: bytes, encoding: str = 'cp437') -> Document:
                                 # HMI 1/1800in -> print columns at 10 CPI,
                                 # the unit .lm/.rm already use (180 = 1 col)
                                 style_fmt[dst_k] = round(hmi / 180)
-                        if entry.get('attrs'):
-                            style_fmt['attrs'] = entry['attrs']
+                        if entry.get('attrs') or _sticky:
+                            # The style's own ON bits, plus every sticky
+                            # attribute still running from an earlier style.
+                            style_fmt['attrs'] = frozenset(
+                                entry.get('attrs') or ()) | style_fmt['sticky_attrs']
                         # Register C5: the style's own declared colour index
                         # (0-15, WSFORMAT's fixed CGA/EGA palette -- same
                         # space as an inline type-1 colour change). `is not