execsql2 2.35.0__py3-none-any.whl → 2.38.0__py3-none-any.whl

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.
Files changed (71) hide show
  1. execsql/api.py +111 -53
  2. execsql/cli/commands/run.py +2 -2
  3. execsql/cli/lint.py +60 -7
  4. execsql/cli/run.py +10 -10
  5. execsql/config.py +7 -0
  6. execsql/data/execsql.conf.template +31 -7
  7. execsql/db/base.py +64 -26
  8. execsql/db/duckdb.py +51 -5
  9. execsql/db/firebird.py +8 -0
  10. execsql/db/mysql.py +8 -3
  11. execsql/db/oracle.py +5 -0
  12. execsql/db/postgres.py +2 -2
  13. execsql/db/sqlite.py +117 -10
  14. execsql/db/sqlserver.py +11 -7
  15. execsql/db/target_types.py +151 -4
  16. execsql/debug/repl.py +4 -3
  17. execsql/exceptions.py +14 -0
  18. execsql/exporters/html.py +26 -42
  19. execsql/exporters/ods.py +93 -21
  20. execsql/exporters/sqlite.py +82 -47
  21. execsql/format.py +12 -4
  22. execsql/gui/desktop.py +2 -1
  23. execsql/gui/tui.py +5 -2
  24. execsql/importers/base.py +9 -7
  25. execsql/importers/json.py +37 -13
  26. execsql/importers/ods.py +22 -2
  27. execsql/lsp/document.py +6 -1
  28. execsql/lsp/hover.py +2 -1
  29. execsql/metacommands/conditions.py +3 -1
  30. execsql/metacommands/connect.py +28 -12
  31. execsql/metacommands/dispatch.py +10 -0
  32. execsql/metacommands/io.py +1 -0
  33. execsql/metacommands/io_export.py +1 -2
  34. execsql/metacommands/io_fileops.py +32 -93
  35. execsql/metacommands/io_import.py +26 -8
  36. execsql/metacommands/io_write.py +3 -1
  37. execsql/metacommands/reference.py +7 -0
  38. execsql/metacommands/script_ext.py +13 -10
  39. execsql/metacommands/system.py +13 -6
  40. execsql/parser.py +14 -37
  41. execsql/script/ast.py +21 -0
  42. execsql/script/control.py +73 -3
  43. execsql/script/executor.py +503 -227
  44. execsql/script/parser.py +102 -52
  45. execsql/script/preparation.py +42 -0
  46. execsql/state.py +23 -0
  47. execsql/types.py +6 -5
  48. execsql/utils/errors.py +24 -6
  49. execsql/utils/fileio.py +342 -134
  50. execsql/utils/sigterm.py +154 -0
  51. execsql/utils/spreadsheet.py +74 -0
  52. execsql/utils/strings.py +13 -7
  53. {execsql2-2.35.0.data → execsql2-2.38.0.data}/data/execsql2_extras/README.md +5 -5
  54. {execsql2-2.35.0.data → execsql2-2.38.0.data}/data/execsql2_extras/example_config_prompt.sql +2 -2
  55. {execsql2-2.35.0.dist-info → execsql2-2.38.0.dist-info}/METADATA +32 -119
  56. {execsql2-2.35.0.dist-info → execsql2-2.38.0.dist-info}/RECORD +71 -68
  57. {execsql2-2.35.0.dist-info → execsql2-2.38.0.dist-info}/WHEEL +1 -1
  58. {execsql2-2.35.0.data → execsql2-2.38.0.data}/data/execsql2_extras/config_settings.sqlite +0 -0
  59. {execsql2-2.35.0.data → execsql2-2.38.0.data}/data/execsql2_extras/make_config_db.sql +0 -0
  60. {execsql2-2.35.0.data → execsql2-2.38.0.data}/data/execsql2_extras/md_compare.sql +0 -0
  61. {execsql2-2.35.0.data → execsql2-2.38.0.data}/data/execsql2_extras/md_glossary.sql +0 -0
  62. {execsql2-2.35.0.data → execsql2-2.38.0.data}/data/execsql2_extras/md_upsert.sql +0 -0
  63. {execsql2-2.35.0.data → execsql2-2.38.0.data}/data/execsql2_extras/pg_compare.sql +0 -0
  64. {execsql2-2.35.0.data → execsql2-2.38.0.data}/data/execsql2_extras/pg_glossary.sql +0 -0
  65. {execsql2-2.35.0.data → execsql2-2.38.0.data}/data/execsql2_extras/script_template.sql +0 -0
  66. {execsql2-2.35.0.data → execsql2-2.38.0.data}/data/execsql2_extras/ss_compare.sql +0 -0
  67. {execsql2-2.35.0.data → execsql2-2.38.0.data}/data/execsql2_extras/ss_glossary.sql +0 -0
  68. {execsql2-2.35.0.data → execsql2-2.38.0.data}/data/execsql2_extras/ss_upsert.sql +0 -0
  69. {execsql2-2.35.0.dist-info → execsql2-2.38.0.dist-info}/entry_points.txt +0 -0
  70. {execsql2-2.35.0.dist-info → execsql2-2.38.0.dist-info}/licenses/LICENSE.txt +0 -0
  71. {execsql2-2.35.0.dist-info → execsql2-2.38.0.dist-info}/licenses/NOTICE +0 -0
@@ -21,7 +21,7 @@ import datetime
21
21
  import math
22
22
  import re
23
23
  from collections.abc import Callable
24
- from decimal import Decimal
24
+ from decimal import Decimal, InvalidOperation
25
25
  from typing import Any, TypeAlias
26
26
 
27
27
  from execsql.exceptions import ErrInfo
@@ -39,7 +39,15 @@ from execsql.types import (
39
39
  DT_TimestampTZ,
40
40
  )
41
41
 
42
- __all__ = ["ColumnKinds", "existing_table_converter", "kind_for_python_type", "kind_for_type_name"]
42
+ __all__ = [
43
+ "decimal_parameter",
44
+ "ColumnKinds",
45
+ "ColumnMetadata",
46
+ "existing_table_converter",
47
+ "kind_for_python_type",
48
+ "kind_for_type_name",
49
+ "kind_and_scale_for_type_name",
50
+ ]
43
51
 
44
52
  #: Lower-cased column name -> the execsql type values for it convert to, or None.
45
53
  #: A string, not ``dict[...]``: on Python 3.10 that object passes
@@ -47,6 +55,17 @@ __all__ = ["ColumnKinds", "existing_table_converter", "kind_for_python_type", "k
47
55
  #: would take it for one.
48
56
  ColumnKinds: TypeAlias = "dict[str, type[DataType] | None]"
49
57
 
58
+
59
+ class ColumnMetadata(dict[str, type[DataType] | None]):
60
+ """A compatible column-kind dictionary carrying known decimal scales."""
61
+
62
+ def __init__(self, columns: dict[str, tuple[type[DataType] | None, int | None]]) -> None:
63
+ super().__init__((name, kind) for name, (kind, _scale) in columns.items())
64
+ self.decimal_scales = {
65
+ name: scale for name, (kind, scale) in columns.items() if kind is DT_Decimal and scale is not None
66
+ }
67
+
68
+
50
69
  _TEXT = {
51
70
  "char",
52
71
  "character",
@@ -162,6 +181,28 @@ def kind_for_type_name(type_name: str, *, oracle: bool = False) -> type[DataType
162
181
  return None
163
182
 
164
183
 
184
+ def kind_and_scale_for_type_name(type_name: str, *, oracle: bool = False) -> tuple[type[DataType] | None, int | None]:
185
+ """Return the existing kind and a declared decimal scale, if known."""
186
+ kind = kind_for_type_name(type_name, oracle=oracle)
187
+ if kind not in (DT_Decimal, DT_Long):
188
+ return kind, None
189
+ name = type_name.strip().lower()
190
+ base = " ".join(_PARAMS.sub(" ", name).replace(" unsigned", " ").replace(" zerofill", " ").split())
191
+ params = _PARAMS.search(name)
192
+ if base not in _DECIMAL or params is None:
193
+ return kind, None
194
+ args = [arg.strip() for arg in params.group(1).split(",")]
195
+ if len(args) == 1:
196
+ return kind, 0
197
+ if len(args) != 2:
198
+ return kind, None
199
+ try:
200
+ scale = int(args[1])
201
+ except ValueError:
202
+ scale = None
203
+ return kind, scale
204
+
205
+
165
206
  def kind_for_python_type(type_code: object, scale: object = None) -> type[DataType] | None:
166
207
  """The execsql type for a column a driver describes by the Python type it returns (pyodbc, firebird-driver).
167
208
 
@@ -201,7 +242,8 @@ Converter = Callable[[Any], Any]
201
242
 
202
243
 
203
244
  def _as_given(value: Any) -> Any:
204
- return value
245
+ # Bind decimal text ourselves before a driver can cast it through DOUBLE.
246
+ return str(value) if isinstance(value, Decimal) and value.is_finite() else value
205
247
 
206
248
 
207
249
  def _whole_number_converter(column: str, table: str, default: Converter) -> Converter:
@@ -255,23 +297,128 @@ def _true_false_as_number(default: Converter) -> Converter:
255
297
  return convert
256
298
 
257
299
 
300
+ def _check_decimal_scale(number: Decimal, value: Any, scale: int | None, column: str, table: str) -> None:
301
+ """Check coefficient digits without quantizing, expanding or using context precision."""
302
+ if scale is None or not number.is_finite() or number.is_zero():
303
+ return
304
+ parts = number.as_tuple()
305
+ exponent = parts.exponent
306
+ assert isinstance(exponent, int)
307
+ discarded = -scale - exponent
308
+ if discarded <= 0:
309
+ return
310
+ if discarded >= len(parts.digits) or any(parts.digits[-discarded:]):
311
+ raise ErrInfo(
312
+ type="error",
313
+ other_msg=(
314
+ f"Can't load {value!r} into column {column} of {table}: the column's scale is {scale}, "
315
+ "and the database would round the value."
316
+ ),
317
+ )
318
+
319
+
320
+ def decimal_parameter(number: Decimal, binding_precision: int | None, column: str, table: str) -> Decimal | str:
321
+ """Bind a finite Decimal exactly within an adapter's native parameter capacity."""
322
+ if binding_precision is None:
323
+ return number
324
+ if number.is_zero():
325
+ return Decimal(0).copy_sign(number)
326
+ parts = number.as_tuple()
327
+ exponent = parts.exponent
328
+ assert isinstance(exponent, int) # Only finite values reach this helper.
329
+ if exponent > 0:
330
+ if number.adjusted() >= binding_precision:
331
+ raise ErrInfo(
332
+ type="error",
333
+ other_msg=(
334
+ f"Can't load {number!r} into column {column} of {table}: "
335
+ f"the database supports at most {binding_precision} decimal digits."
336
+ ),
337
+ )
338
+ # Expansion is bounded by the adapter, never by the input exponent.
339
+ return Decimal(format(number, "f"))
340
+ native_width = len(parts.digits)
341
+ if -exponent > native_width:
342
+ native_width = -exponent + 1
343
+ if native_width > binding_precision:
344
+ # Avoid a driver's silent Decimal-to-double fallback. Fixed text
345
+ # also avoids exponent parsing at the maximum supported scale.
346
+ if exponent >= -binding_precision:
347
+ return format(number, "f")
348
+ return str(number)
349
+ return number
350
+
351
+
352
+ def _exact_decimal_converter(
353
+ default: Converter,
354
+ column: str,
355
+ table: str,
356
+ binding_precision: int | None,
357
+ scale: int | None,
358
+ ) -> Converter:
359
+ """Retain finite exact source numbers for an existing decimal column.
360
+
361
+ Floats and non-numeric values keep the scanned conversion; converting a
362
+ float here cannot recover source digits already lost by its reader.
363
+ """
364
+
365
+ def convert(value: Any) -> Any:
366
+ if isinstance(value, bool):
367
+ return default(value)
368
+ if isinstance(value, int):
369
+ _check_decimal_scale(Decimal(value), value, scale, column, table)
370
+ return value
371
+ if isinstance(value, Decimal) and value.is_finite():
372
+ _check_decimal_scale(value, value, scale, column, table)
373
+ return decimal_parameter(value, binding_precision, column, table)
374
+ if isinstance(value, str) and _NUMBER_TEXT.fullmatch(value):
375
+ try:
376
+ number = Decimal(value)
377
+ except InvalidOperation:
378
+ return default(value)
379
+ if number.is_finite():
380
+ _check_decimal_scale(number, value, scale, column, table)
381
+ return decimal_parameter(number, binding_precision, column, table)
382
+ return default(value)
383
+
384
+ return convert
385
+
386
+
258
387
  def existing_table_converter(
259
388
  kind: type[DataType] | None,
260
389
  scanned: type[DataType],
261
390
  default: Converter,
262
391
  column: str,
263
392
  table: str,
393
+ *,
394
+ decimal_binding_precision: int | None = None,
395
+ decimal_scale: int | None = None,
264
396
  ) -> Converter:
265
397
  """How a value for *column* of existing table *table* is converted.
266
398
 
267
399
  *kind* is the column's type (``None`` if execsql has none that fits),
268
400
  *scanned* the type the source column scanned as, and *default* the
269
401
  conversion to that scanned type, which a new table's values get.
402
+ *decimal_binding_precision* bounds native Decimal parameter conversion
403
+ for drivers that need it. *decimal_scale* rejects exact source values that
404
+ need rounding; None leaves the destination scale to the database.
270
405
  """
271
406
  if kind is DT_Text:
272
407
  return _as_given
273
408
  if kind in (DT_Long, DT_Decimal, DT_Float) and scanned is DT_Boolean:
274
- return _true_false_as_number(default)
409
+ boolean_converter = _true_false_as_number(default)
410
+ if kind is DT_Decimal and decimal_scale is not None:
411
+
412
+ def checked_boolean(value: Any) -> Any:
413
+ result = boolean_converter(value)
414
+ if isinstance(result, (int, Decimal)):
415
+ _check_decimal_scale(Decimal(result), value, decimal_scale, column, table)
416
+ return result
417
+
418
+ return checked_boolean
419
+ return boolean_converter
275
420
  if kind is DT_Long:
276
421
  return _whole_number_converter(column, table, default)
422
+ if kind is DT_Decimal:
423
+ return _exact_decimal_converter(default, column, table, decimal_binding_precision, decimal_scale)
277
424
  return default
execsql/debug/repl.py CHANGED
@@ -6,8 +6,8 @@ Input runs as if it were written where the script paused, through
6
6
  :mod:`execsql.interactive`: SQL (ending with ``;``, results shown as tables),
7
7
  metacommands and blocks, in the paused script's variable scope and
8
8
  transaction. A ``SUB`` typed here is still set
9
- when the script resumes; a ``BREAK`` inside a LOOP leaves the loop; ``HALT``
10
- ends the run. An error ends only the input that caused it.
9
+ when the script resumes; a ``BREAK`` leaves the LOOP, SCRIPT or included file
10
+ the script paused in; ``HALT`` ends the run. An error ends only the input that caused it.
11
11
 
12
12
  Commands to the REPL itself start with ``.``: ``.continue`` and ``.next``
13
13
  resume the script, ``.where`` and ``.stack`` show where it paused, ``.quit``
@@ -27,6 +27,7 @@ from pathlib import Path
27
27
  from typing import Any
28
28
 
29
29
  import execsql.state as _state
30
+ from execsql.exceptions import ScriptExit
30
31
  from execsql.interactive import Prompt, common_dot_command, read_eval_loop, unknown_dot_command
31
32
  from execsql.utils.color import color_disabled_by_env
32
33
 
@@ -237,7 +238,7 @@ def _handle_dot_command(line: str) -> bool:
237
238
  if command in ("continue", "c"):
238
239
  return False
239
240
  if command in ("abort", "q", "quit"):
240
- raise SystemExit(1)
241
+ raise ScriptExit(1)
241
242
  if command in ("next", "n"):
242
243
  _enable_step_mode()
243
244
  return False
execsql/exceptions.py CHANGED
@@ -21,6 +21,7 @@ can import from a single location. Notable exceptions:
21
21
  — spreadsheet I/O failures.
22
22
  - :class:`ConsoleUIError` — GUI console errors.
23
23
  - :class:`CondParserError` / :class:`NumericParserError` — parser failures.
24
+ - :class:`ScriptExit` — execsql ending the run on purpose (a ``SystemExit``).
24
25
 
25
26
  Note:
26
27
  The ``execsql`` top-level package re-exports a *separate*
@@ -49,6 +50,7 @@ __all__ = [
49
50
  "ConsoleUIError",
50
51
  "CondParserError",
51
52
  "NumericParserError",
53
+ "ScriptExit",
52
54
  ]
53
55
 
54
56
 
@@ -272,3 +274,15 @@ class CondParserError(ExecSqlError):
272
274
 
273
275
  class NumericParserError(ExecSqlError):
274
276
  """Raised for numeric-expression parse errors."""
277
+
278
+
279
+ class ScriptExit(SystemExit):
280
+ """execsql ending the run on purpose.
281
+
282
+ Raised by ``exit_now()`` (``HALT``, an error that halts the run, a
283
+ cancelled ``PROMPT``), by a cancelled dialog, and by ``.quit`` / ``.abort``
284
+ at a ``BREAKPOINT`` prompt. A ``SystemExit``, so everything that catches
285
+ one still does. A plain ``SystemExit`` -- from a signal handler, execsql's
286
+ own SIGTERM handler, or quitting the Textual console -- is an interrupt:
287
+ the statement it stops is rolled back like a failed one.
288
+ """
execsql/exporters/html.py CHANGED
@@ -28,6 +28,26 @@ from execsql.utils.fileio import filewriter_close
28
28
  __all__ = ["export_html", "export_cgi_html", "write_query_to_html", "write_query_to_cgi_html"]
29
29
 
30
30
 
31
+ def _write_table(f: Any, hdrs: list[str], rows: Any, desc: str | None) -> None:
32
+ """Write the table fragment shared by standalone and CGI HTML exports."""
33
+ f.write("<table>\n")
34
+ if desc is not None:
35
+ f.write(f"<caption>{desc}</caption>\n")
36
+ f.write("<thead><tr>")
37
+ for h in hdrs:
38
+ f.write(f"<th>{html_mod.escape(str(h))}</th>")
39
+ f.write("</tr></thead>\n<tbody>\n")
40
+ for r in rows:
41
+ f.write("<tr>")
42
+ for v in r:
43
+ # `if v is not None`, not `if v`: 0, False, and an empty
44
+ # string are all falsy, and a zero measurement silently became
45
+ # an empty cell.
46
+ f.write(f"<td>{html_mod.escape(str(v)) if v is not None else ''}</td>")
47
+ f.write("</tr>\n")
48
+ f.write("</tbody>\n</table>\n")
49
+
50
+
31
51
  def export_html(
32
52
  outfile: str,
33
53
  hdrs: list[str],
@@ -40,24 +60,6 @@ def export_html(
40
60
  """Write a complete HTML document containing a data table to a file or ZIP archive."""
41
61
  conf = _state.conf
42
62
 
43
- def write_table(f):
44
- f.write("<table>\n")
45
- if desc is not None:
46
- f.write(f"<caption>{desc}</caption>\n")
47
- f.write("<thead><tr>")
48
- for h in hdrs:
49
- f.write(f"<th>{html_mod.escape(str(h))}</th>")
50
- f.write("</tr></thead>\n<tbody>\n")
51
- for r in rows:
52
- f.write("<tr>")
53
- for v in r:
54
- # `if v is not None`, not `if v`: 0, False, and an empty
55
- # string are all falsy, and a zero measurement silently became
56
- # an empty cell.
57
- f.write(f"<td>{html_mod.escape(str(v)) if v is not None else ''}</td>")
58
- f.write("</tr>\n")
59
- f.write("</tbody>\n</table>\n")
60
-
61
63
  script, lno = current_script_line()
62
64
  # If not append, write a complete HTML document with header and table.
63
65
  # If append and the file does not exist, write just the table.
@@ -112,7 +114,7 @@ def export_html(
112
114
  f.write("tbody tr { border-bottom: 1px dotted #814324; }\n")
113
115
  f.write("</style>")
114
116
  f.write("\n</head>\n<body>\n")
115
- write_table(f)
117
+ _write_table(f, hdrs, rows, desc)
116
118
  f.write("</body>\n</html>\n")
117
119
  finally:
118
120
  if outfile.lower() != "stdout":
@@ -120,14 +122,14 @@ def export_html(
120
122
  elif not zipfile and append:
121
123
  if outfile.lower() == "stdout":
122
124
  f = sys.stdout
123
- write_table(f)
125
+ _write_table(f, hdrs, rows, desc)
124
126
  elif not Path(outfile).is_file():
125
127
  from execsql.utils.fileio import EncodedFile
126
128
 
127
129
  ef = EncodedFile(outfile, conf.output_encoding)
128
130
  f = ef.open("wt")
129
131
  try:
130
- write_table(f)
132
+ _write_table(f, hdrs, rows, desc)
131
133
  finally:
132
134
  f.close()
133
135
  else:
@@ -152,7 +154,7 @@ def export_html(
152
154
  else:
153
155
  t.write(line)
154
156
  t.write("\n")
155
- write_table(t)
157
+ _write_table(t, hdrs, rows, desc)
156
158
  t.write(remainder)
157
159
  for line in f:
158
160
  t.write(line)
@@ -183,24 +185,6 @@ def export_cgi_html(
183
185
  """Write a CGI-style HTML fragment (Content-Type header + table) to a file or ZIP archive."""
184
186
  conf = _state.conf
185
187
 
186
- def write_table(f):
187
- f.write("<table>\n")
188
- if desc is not None:
189
- f.write(f"<caption>{desc}</caption>\n")
190
- f.write("<thead><tr>")
191
- for h in hdrs:
192
- f.write(f"<th>{html_mod.escape(str(h))}</th>")
193
- f.write("</tr></thead>\n<tbody>\n")
194
- for r in rows:
195
- f.write("<tr>")
196
- for v in r:
197
- # `if v is not None`, not `if v`: 0, False, and an empty
198
- # string are all falsy, and a zero measurement silently became
199
- # an empty cell.
200
- f.write(f"<td>{html_mod.escape(str(v)) if v is not None else ''}</td>")
201
- f.write("</tr>\n")
202
- f.write("</tbody>\n</table>\n")
203
-
204
188
  script, lno = current_script_line()
205
189
  if zipfile or not append or (append and not Path(outfile).is_file()):
206
190
  if zipfile is None:
@@ -216,7 +200,7 @@ def export_cgi_html(
216
200
  f = ZipWriter(zipfile, outfile, append)
217
201
  try:
218
202
  f.write("Content-Type: text/html\n\n")
219
- write_table(f)
203
+ _write_table(f, hdrs, rows, desc)
220
204
  finally:
221
205
  if outfile.lower() != "stdout":
222
206
  f.close()
@@ -229,7 +213,7 @@ def export_cgi_html(
229
213
  ef = EncodedFile(outfile, conf.output_encoding)
230
214
  f = ef.open("a")
231
215
  try:
232
- write_table(f)
216
+ _write_table(f, hdrs, rows, desc)
233
217
  finally:
234
218
  if outfile.lower() != "stdout":
235
219
  f.close()
execsql/exporters/ods.py CHANGED
@@ -24,6 +24,7 @@ from execsql.script import current_script_line
24
24
  from execsql.utils.errors import exception_desc, fatal_error, install_hint
25
25
  from execsql.utils.fileio import filewriter_close
26
26
  from execsql.utils.strings import unquoted
27
+ from execsql.utils.spreadsheet import CellTextBudget, SheetBudget
27
28
 
28
29
  __all__ = ["OdsFile", "export_ods", "write_query_to_ods", "write_queries_to_ods"]
29
30
 
@@ -179,26 +180,36 @@ class OdsFile:
179
180
  return sheet
180
181
  return None
181
182
 
182
- def sheet_data(self, sheetname: Any, junk_header_rows: int = 0) -> list:
183
- """Return all row data from the named sheet, optionally skipping leading junk rows."""
183
+ def sheet_data(self, sheetname: Any, junk_header_rows: int = 0, *, budget: SheetBudget | None = None) -> list:
184
+ """Read a sheet after SKIP, checking optional budgets before expanding rows or padding."""
184
185
  sheet = self.sheet_named(sheetname)
185
186
  if not sheet:
186
187
  raise OdsFileError(f"There is no sheet named {sheetname}")
187
188
 
188
189
  rows: list[list | None] = []
189
190
  pending_empty = 0
191
+ skip_rows = max(0, junk_header_rows)
190
192
  for sheetrow in sheet.getElementsByType(of.table.TableRow):
191
- data = _row_values(sheetrow)
192
- repeat = int(sheetrow.getAttribute("numberrowsrepeated") or 1)
193
+ repeat = _expansion_count(sheetrow, "numberrowsrepeated")
194
+ skipped = min(skip_rows, repeat)
195
+ skip_rows -= skipped
196
+ repeat -= skipped
197
+ if repeat == 0:
198
+ _validate_expansion_counts(sheetrow)
199
+ continue
200
+ data = _row_values(sheetrow, budget, row_repeat=repeat, pending_rows=pending_empty)
193
201
  if not data:
194
202
  # Empty rows are kept only when data follows them, so the
195
203
  # filler rows a spreadsheet writes below the data are dropped.
196
204
  pending_empty += repeat
197
205
  continue
206
+ if budget is not None:
207
+ budget.add_row(data, repeat, pending_empty)
198
208
  rows.extend(None for _ in range(pending_empty))
199
209
  pending_empty = 0
200
210
  rows.extend(list(data) for _ in range(repeat))
201
- rows = rows[junk_header_rows:]
211
+ if junk_header_rows < 0:
212
+ rows = rows[junk_header_rows:]
202
213
  # Every row is as wide as the widest, so a blank trailing header is
203
214
  # still seen, and an empty row inside the data is a row of NULLs that
204
215
  # EMPTY_ROWS keeps or drops.
@@ -309,20 +320,55 @@ class OdsFile:
309
320
  self.wbk = None
310
321
 
311
322
 
312
- def _paragraph_text(node: Any) -> str:
323
+ def _expansion_count(node: Any, attribute: str) -> int:
324
+ """Read a non-negative ODS repeat/span/space count without expanding it."""
325
+ value = node.getAttribute(attribute) or "1"
326
+ try:
327
+ count = int(value)
328
+ if count < 0:
329
+ raise ValueError("negative count")
330
+ except (ValueError, TypeError) as exc:
331
+ shown = value[:120] + "..." if isinstance(value, str) and len(value) > 120 else value
332
+ raise ErrInfo(type="error", other_msg=f"Invalid ODS expansion count {shown!r} for {attribute}.") from exc
333
+ return count
334
+
335
+
336
+ def _validate_expansion_counts(node: Any) -> None:
337
+ """Validate skipped content without allocating its repeated rows or text."""
338
+ qname = getattr(node, "qname", None)
339
+ if qname == (of.table.TABLENS, "table-row"):
340
+ _expansion_count(node, "numberrowsrepeated")
341
+ elif qname in ((of.table.TABLENS, "table-cell"), (of.table.TABLENS, "covered-table-cell")):
342
+ _expansion_count(node, "numbercolumnsrepeated")
343
+ if qname == (of.table.TABLENS, "table-cell"):
344
+ _expansion_count(node, "numbercolumnsspanned")
345
+ _expansion_count(node, "numberrowsspanned")
346
+ elif qname == (of.text.TEXTNS, "s"):
347
+ _expansion_count(node, "c")
348
+ for child in getattr(node, "childNodes", ()):
349
+ _validate_expansion_counts(child)
350
+
351
+
352
+ def _paragraph_text(node: Any, budget: CellTextBudget | None = None) -> str:
313
353
  """Return the text of a ``<text:p>``, expanding ODF's space, tab and line-break elements."""
314
354
  parts: list[str] = []
315
355
  for child in node.childNodes:
316
356
  if child.nodeType == child.TEXT_NODE:
317
- parts.append(str(child))
357
+ value = str(child)
358
+ if budget is not None:
359
+ budget.add(len(value))
318
360
  elif child.qname == (of.text.TEXTNS, "s"):
319
- parts.append(" " * int(child.getAttribute("c") or 1))
320
- elif child.qname == (of.text.TEXTNS, "tab"):
321
- parts.append("\t")
322
- elif child.qname == (of.text.TEXTNS, "line-break"):
323
- parts.append("\n")
361
+ count = _expansion_count(child, "c")
362
+ if budget is not None:
363
+ budget.add(count)
364
+ value = " " * count
365
+ elif child.qname in ((of.text.TEXTNS, "tab"), (of.text.TEXTNS, "line-break")):
366
+ if budget is not None:
367
+ budget.add(1)
368
+ value = "\t" if child.qname == (of.text.TEXTNS, "tab") else "\n"
324
369
  else:
325
- parts.append(_paragraph_text(child))
370
+ value = _paragraph_text(child, budget)
371
+ parts.append(value)
326
372
  return "".join(parts)
327
373
 
328
374
 
@@ -336,7 +382,7 @@ def _time_text(duration: str) -> str:
336
382
  return f"{int(hours):02d}:{int(minutes):02d}:{int(seconds):02d}{fraction}"
337
383
 
338
384
 
339
- def _cell_value(cell: Any) -> str | None:
385
+ def _cell_value(cell: Any, budget: CellTextBudget | None = None) -> str | None:
340
386
  """Return a cell's stored value as text, or ``None`` for an empty cell.
341
387
 
342
388
  The value comes from the cell's ``office:value-type`` and value
@@ -361,25 +407,39 @@ def _cell_value(cell: Any) -> str | None:
361
407
  else:
362
408
  value = None
363
409
  if value:
410
+ if budget is not None:
411
+ budget.add(len(value))
364
412
  return value
365
413
  if vtype == "string":
366
414
  value = cell.getAttribute("stringvalue")
367
415
  if value:
416
+ if budget is not None:
417
+ budget.add(len(value))
368
418
  return value
369
419
  # A string cell, a cell with no type, or a typed cell missing its value
370
420
  # attribute: the paragraphs are the value, one line each.
371
- paragraphs = [
372
- _paragraph_text(child) for child in cell.childNodes if getattr(child, "qname", None) == (of.text.TEXTNS, "p")
373
- ]
421
+ paragraphs: list[str] = []
422
+ for child in cell.childNodes:
423
+ if getattr(child, "qname", None) == (of.text.TEXTNS, "p"):
424
+ if paragraphs and budget is not None:
425
+ budget.add(1)
426
+ paragraphs.append(_paragraph_text(child, budget))
374
427
  text = "\n".join(paragraphs)
375
428
  return text or None
376
429
 
377
430
 
378
- def _row_values(sheetrow: Any) -> list:
431
+ def _row_values(
432
+ sheetrow: Any,
433
+ budget: SheetBudget | None = None,
434
+ *,
435
+ row_repeat: int = 1,
436
+ pending_rows: int = 0,
437
+ ) -> list:
379
438
  """Return one value per column of a ``<table:table-row>``, without trailing empty cells."""
380
439
  values: list = []
381
440
  pending_empty = 0
382
441
  covered_by_span = 0
442
+ row_text = 0
383
443
  for cell in sheetrow.childNodes:
384
444
  qname = getattr(cell, "qname", None)
385
445
  if qname == (of.table.TABLENS, "table-cell"):
@@ -388,7 +448,7 @@ def _row_values(sheetrow: Any) -> list:
388
448
  covered = True
389
449
  else:
390
450
  continue
391
- repeat = int(cell.getAttribute("numbercolumnsrepeated") or 1)
451
+ repeat = _expansion_count(cell, "numbercolumnsrepeated")
392
452
  if covered:
393
453
  # A horizontal merge repeats its value across the columns it
394
454
  # covers (below); covered cells left over belong to a merge from
@@ -398,11 +458,20 @@ def _row_values(sheetrow: Any) -> list:
398
458
  repeat -= skip
399
459
  value = None
400
460
  else:
401
- value = _cell_value(cell)
402
- spanned = int(cell.getAttribute("numbercolumnsspanned") or 1)
461
+ spanned = _expansion_count(cell, "numbercolumnsspanned")
462
+ _expansion_count(cell, "numberrowsspanned")
403
463
  if spanned > 1:
404
464
  covered_by_span = spanned - 1
405
465
  repeat *= spanned
466
+ if repeat == 0:
467
+ _validate_expansion_counts(cell)
468
+ continue
469
+ text_budget = (
470
+ CellTextBudget(budget, budget.text_chars + row_text * row_repeat, repeat * row_repeat)
471
+ if budget is not None
472
+ else None
473
+ )
474
+ value = _cell_value(cell, text_budget)
406
475
  if repeat <= 0:
407
476
  continue
408
477
  if value is None:
@@ -410,6 +479,9 @@ def _row_values(sheetrow: Any) -> list:
410
479
  # repeated across the whole sheet width costs nothing.
411
480
  pending_empty += repeat
412
481
  continue
482
+ if budget is not None:
483
+ budget.check_row(len(values) + pending_empty + repeat, row_repeat, pending_rows)
484
+ row_text += len(value) * repeat
413
485
  values.extend([None] * pending_empty)
414
486
  pending_empty = 0
415
487
  values.extend([value] * repeat)