sdb-reader 0.10.0__tar.gz → 0.11.0__tar.gz

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 (44) hide show
  1. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/CHANGELOG.md +43 -0
  2. {sdb_reader-0.10.0/src/sdb_reader.egg-info → sdb_reader-0.11.0}/PKG-INFO +1 -1
  3. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/pyproject.toml +1 -1
  4. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/src/sdb_reader/__init__.py +6 -2
  5. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/src/sdb_reader/compliance.py +162 -1
  6. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/src/sdb_reader/messages.py +5 -0
  7. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/src/sdb_reader/messages_data/de.toml +5 -0
  8. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/src/sdb_reader/messages_data/en.toml +5 -0
  9. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/src/sdb_reader/parser.py +112 -5
  10. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/src/sdb_reader/read_record.schema.json +24 -0
  11. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/src/sdb_reader/schema.py +3 -0
  12. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/src/sdb_reader/serialise.py +9 -1
  13. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/src/sdb_reader/statements.py +51 -1
  14. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/src/sdb_reader/versions.py +1 -1
  15. {sdb_reader-0.10.0 → sdb_reader-0.11.0/src/sdb_reader.egg-info}/PKG-INFO +1 -1
  16. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/LICENSE +0 -0
  17. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/MANIFEST.in +0 -0
  18. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/NOTICE +0 -0
  19. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/README.md +0 -0
  20. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/setup.cfg +0 -0
  21. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/src/sdb_reader/ask.py +0 -0
  22. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/src/sdb_reader/ask_record.schema.json +0 -0
  23. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/src/sdb_reader/cli.py +0 -0
  24. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/src/sdb_reader/clp_statements_de.json +0 -0
  25. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/src/sdb_reader/current.py +0 -0
  26. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/src/sdb_reader/diff.py +0 -0
  27. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/src/sdb_reader/embedded.py +0 -0
  28. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/src/sdb_reader/extraction.py +0 -0
  29. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/src/sdb_reader/layout.py +0 -0
  30. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/src/sdb_reader/line_breaks.py +0 -0
  31. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/src/sdb_reader/model.py +0 -0
  32. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/src/sdb_reader/pictograms.py +0 -0
  33. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/src/sdb_reader/presets.py +0 -0
  34. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/src/sdb_reader/presets.toml +0 -0
  35. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/src/sdb_reader/py.typed +0 -0
  36. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/src/sdb_reader/reading.py +0 -0
  37. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/src/sdb_reader/scan.py +0 -0
  38. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/src/sdb_reader/show.py +0 -0
  39. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/src/sdb_reader/verification.py +0 -0
  40. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/src/sdb_reader.egg-info/SOURCES.txt +0 -0
  41. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/src/sdb_reader.egg-info/dependency_links.txt +0 -0
  42. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/src/sdb_reader.egg-info/entry_points.txt +0 -0
  43. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/src/sdb_reader.egg-info/requires.txt +0 -0
  44. {sdb_reader-0.10.0 → sdb_reader-0.11.0}/src/sdb_reader.egg-info/top_level.txt +0 -0
@@ -5,6 +5,49 @@ Versions follow the tags of this repository.
5
5
 
6
6
  ## Unreleased
7
7
 
8
+ ## 0.11.0 – 2026-11-21
9
+
10
+ ### Breaking
11
+
12
+ - **`PARSER_VERSION` is 20** (0.10.0: 19): findings are part of the reading.
13
+ `EXTRACTION_VERSION` stays 2; readings kept from before should be made
14
+ again from the stored texts.
15
+
16
+ ### Added
17
+
18
+ - **Whether the label prints a statement's code.** Each statement of the
19
+ label says under `code_source` whether the sheet prints its code
20
+ (`printed`) or only its sentence, to which the parser assigned the code
21
+ (`from_wording`), and `hazard_statements.codes_from_wording` lists the
22
+ assigned codes. The code lists stay as they are. `as_printed(text,
23
+ codes_from_wording)` gives a statement text without the assigned codes,
24
+ as the sheet prints it (SDB-174).
25
+ - **A hazard statement of the classification that the label does not
26
+ carry.** Where section 2.1 names an H statement for the classification and
27
+ the label carries neither its code nor its wording, the sheet gets the
28
+ finding `hazard-statement-not-in-label` with the code. Not required are H400
29
+ beside H410 and H318 beside H314 (CLP Annex III Part 1) and H304 on an
30
+ aerosol (Annex I 1.3.3); not where the label carries no H statement at all,
31
+ which `label-without-hazard-statement` says (SDB-173).
32
+ - **Findings for the label of a classified product.** Where section 2
33
+ classifies the product under CLP and the label (2.2) is not found
34
+ (`label-not-found`), carries no H statement
35
+ (`label-without-hazard-statement`) or no P statement
36
+ (`label-without-precautionary-statement`), the sheet gets a finding. The
37
+ reading stays as the sheet prints it; nothing is added. Each finding says
38
+ "not found": the sheet may lack it, or the parser may have missed it. Not
39
+ for an article, a classification under the old law, or a label under the
40
+ directives before CLP on a sheet dated before 1 June 2015; the parser notes
41
+ such a label (`label-before-clp`) (SDB-169 to SDB-172).
42
+
43
+ ### Fixed
44
+
45
+ - **A section heading numbered with a colon starts its section.** "3:
46
+ Zusammensetzung / Angaben zu den Bestandteilen" is the heading of section 3
47
+ as "3. Zusammensetzung …" is, when the title fits the number; before, section
48
+ 2 ran on into section 3, and an ingredient's classification counted as the
49
+ product's (SDB-175).
50
+
8
51
  ## 0.10.0 – 2026-11-20
9
52
 
10
53
  ### Breaking
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.4
2
2
  Name: sdb-reader
3
- Version: 0.10.0
3
+ Version: 0.11.0
4
4
  Summary: Reads German safety data sheets (Sicherheitsdatenblätter) under REACH and CLP, without a language model
5
5
  Author-email: "Förde-Baumarkt Holger Claussen GmbH & Co. KG" <service@foerde-baumarkt.de>
6
6
  License-Expression: AGPL-3.0-or-later
@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
4
4
 
5
5
  [project]
6
6
  name = "sdb-reader"
7
- version = "0.10.0"
7
+ version = "0.11.0"
8
8
  description = "Reads German safety data sheets (Sicherheitsdatenblätter) under REACH and CLP, without a language model"
9
9
  readme = "README.md"
10
10
  authors = [{ name = "Förde-Baumarkt Holger Claussen GmbH & Co. KG", email = "service@foerde-baumarkt.de" }]
@@ -28,8 +28,10 @@ with any release, the parts that ask a language model included.
28
28
  - Judging a sheet: ``sheet_findings`` and the checks it is made of,
29
29
  ``classification`` and ``requires_sheet``.
30
30
  - Statements and pictograms: ``label_statements`` with the regulation's
31
- wording (``canonical_text``), the statement texts tidied and rendered, the
32
- pictograms that follow from H statements.
31
+ wording (``canonical_text``) and whether the label prints each code, the
32
+ statement texts tidied and rendered, ``as_printed`` for the texts without
33
+ the codes the label does not print, the pictograms that follow from H
34
+ statements.
33
35
  - Embedded data: ``read_embedded``, ``parse_embedded`` and
34
36
  ``embedded_differences``.
35
37
  - Messages: what the library notes about a file is a ``Message`` with a
@@ -90,6 +92,7 @@ from .serialise import (
90
92
  )
91
93
  from .statements import (
92
94
  StatementWording,
95
+ as_printed,
93
96
  canonical_text,
94
97
  label_statements,
95
98
  normalize_statements,
@@ -125,6 +128,7 @@ __all__ = [
125
128
  "StatedClassification",
126
129
  "StatementWording",
127
130
  "Verification",
131
+ "as_printed",
128
132
  "canonical_text",
129
133
  "classification",
130
134
  "contradicts_label",
@@ -18,11 +18,22 @@ from .messages import Message
18
18
  from .parser import (
19
19
  CLASSIFIED_BEFORE_CLP,
20
20
  EUH_WORDING_OUTSIDE_LABEL,
21
+ LABEL_BEFORE_CLP,
22
+ LABEL_HEADING_PATTERN,
21
23
  LABEL_UNDETERMINED,
24
+ NO_LABEL_SECTION,
22
25
  SheetPart,
26
+ classification_hazard_codes,
23
27
  )
24
28
  from .schema import HazardStatements, SafetyDataSheet, Section, StatedClassification
25
- from .statements import SHEET_ON_REQUEST, deviates_from_regulation, label_statements
29
+ from .statements import (
30
+ SHEET_ON_REQUEST,
31
+ _catalogue,
32
+ _comparable,
33
+ base_code,
34
+ deviates_from_regulation,
35
+ label_statements,
36
+ )
26
37
 
27
38
  # Annex II of REACH in the version of Regulation (EU) 2020/878 applies from
28
39
  # 1 January 2021; sheets not conforming to it could be supplied until
@@ -253,6 +264,152 @@ def contradicts_label(sheet: SafetyDataSheet) -> bool:
253
264
  )
254
265
 
255
266
 
267
+ def label_not_found(
268
+ sheet: SafetyDataSheet, parser_warnings: Iterable[Message]
269
+ ) -> list[Message]:
270
+ """What the label of a product section 2 classifies should carry and the
271
+ parser did not find in it (SDB-169 to SDB-172).
272
+
273
+ The reading stays as it is: these are findings beside it, and each says
274
+ "not found", which is true whether the sheet lacks it or the parser missed
275
+ it. Only for a product classified under CLP (SDB-75); not under the old
276
+ law (SDB-73), not for a label under the old directives on a sheet dated
277
+ before 1 June 2015, and not for an article, which is not labelled under
278
+ CLP (SDB-142, SDB-154). A label counts as found where the parser found its
279
+ block (SDB-4), or, without one, read any element of a label in section 2.
280
+
281
+ >>> from sdb_reader import HazardStatements
282
+ >>> def sheet(label, stated=StatedClassification.CLASSIFIED):
283
+ ... return SafetyDataSheet(
284
+ ... None, None, None, None, "de", hazard_statements=label,
285
+ ... stated_classification=stated,
286
+ ... )
287
+ >>> [m.code for m in label_not_found(sheet(HazardStatements()),
288
+ ... [Message(NO_LABEL_SECTION)])]
289
+ ['label-not-found']
290
+ >>> [m.code for m in label_not_found(
291
+ ... sheet(HazardStatements(signal_word="Achtung", precautionary_codes=["P210"])),
292
+ ... [])]
293
+ ['label-without-hazard-statement']
294
+ >>> label_not_found(sheet(HazardStatements(), StatedClassification.NOT_STATED), [])
295
+ []
296
+ """
297
+ noted = {warning.code for warning in parser_warnings}
298
+ if not _label_is_judged(sheet, noted):
299
+ return []
300
+ label = sheet.hazard_statements
301
+ read_any = bool(
302
+ label.hazard_codes
303
+ or label.precautionary_codes
304
+ or label.supplemental_codes
305
+ or label.signal_word
306
+ or label.pictograms
307
+ )
308
+ if NO_LABEL_SECTION in noted and not read_any:
309
+ return [Message("label-not-found")]
310
+ findings: list[Message] = []
311
+ if not label.hazard_codes:
312
+ findings.append(Message("label-without-hazard-statement"))
313
+ if not label.precautionary_codes:
314
+ findings.append(Message("label-without-precautionary-statement"))
315
+ return findings
316
+
317
+
318
+ def _label_is_judged(sheet: SafetyDataSheet, noted: Collection[str]) -> bool:
319
+ """Whether the label findings apply to *sheet* (SDB-170): a product
320
+ classified under CLP, no article, no classification under the old law, and
321
+ no label under the old directives on a sheet dated before 1 June 2015.
322
+ *noted* are the codes of the parser's warnings."""
323
+ if sheet.stated_classification is not StatedClassification.CLASSIFIED:
324
+ return False
325
+ if sheet.stated_article or CLASSIFIED_BEFORE_CLP in noted:
326
+ return False
327
+ # A mixture could be labelled under the old directives until 1 June 2015.
328
+ dated = sheet.revision_date or sheet.issue_date
329
+ return not (
330
+ LABEL_BEFORE_CLP in noted and dated is not None and dated < CLP_FOR_MIXTURES
331
+ )
332
+
333
+
334
+ # The order of precedence Annex III Part 1 sets under Article 27 (consolidated
335
+ # version of 1 July 2026): a statement the label may leave out beside another.
336
+ _LEFT_OUT_BESIDE = {"H400": "H410", "H318": "H314"}
337
+ # An aerosol: its class in 2.1 or one of its statements (Annex I 2.3).
338
+ _AEROSOL_CLASS = re.compile(r"\bAerosol\s?[123]\b")
339
+ _AEROSOL_STATEMENTS = frozenset({"H222", "H223", "H229"})
340
+
341
+
342
+ def _not_required(
343
+ code: str, on_label: set[str], classified: set[str], hazards: str
344
+ ) -> bool:
345
+ """Whether the label may go without *code* although 2.1 names it (SDB-173).
346
+
347
+ Beside H410 or H314 the label may leave out H400 or H318 (CLP Annex III
348
+ Part 1 under Article 27). A product placed on the market as an aerosol need
349
+ not be labelled for the aspiration hazard, H304 (Annex I 1.3.3).
350
+ """
351
+ if _LEFT_OUT_BESIDE.get(code) in on_label:
352
+ return True
353
+ return code == "H304" and bool(
354
+ _AEROSOL_CLASS.search(hazards) or (classified | on_label) & _AEROSOL_STATEMENTS
355
+ )
356
+
357
+
358
+ def hazard_statements_not_in_label(
359
+ sheet: SafetyDataSheet, parser_warnings: Iterable[Message]
360
+ ) -> list[Message]:
361
+ r"""Each H statement of the classification in 2.1 that the label does not
362
+ carry (SDB-173), with its code as ``statement``.
363
+
364
+ Only where the label carries at least one H statement -- without any,
365
+ ``label-without-hazard-statement`` says so (SDB-171) -- and with the
366
+ exceptions of SDB-170. A code counts as carried where the label carries the
367
+ same statement, H360 and H361 whatever their letters, and H400 and H318 also
368
+ where the label carries H410 or H314 (CLP Article 27 with Annex III Part 1).
369
+
370
+ >>> from sdb_reader import HazardStatements
371
+ >>> section = (
372
+ ... "2.1 Einstufung des Stoffs oder Gemischs\n"
373
+ ... "Aerosol 3, H229\nAquatic Acute 1, H400\nAquatic Chronic 1, H410\n"
374
+ ... "2.2 Kennzeichnungselemente\n"
375
+ ... )
376
+ >>> sheet = SafetyDataSheet(
377
+ ... None, None, None, None, "de",
378
+ ... hazard_statements=HazardStatements(hazard_codes=["H410"]),
379
+ ... stated_classification=StatedClassification.CLASSIFIED,
380
+ ... section_text={Section.HAZARDS: section},
381
+ ... )
382
+ >>> [m.params["statement"] for m in hazard_statements_not_in_label(sheet, [])]
383
+ ['H229']
384
+ """
385
+ noted = {warning.code for warning in parser_warnings}
386
+ label = sheet.hazard_statements
387
+ if not label.hazard_codes or not _label_is_judged(sheet, noted):
388
+ return []
389
+ hazards = sheet.section_text.get(Section.HAZARDS, "")
390
+ heading = LABEL_HEADING_PATTERN.search(hazards)
391
+ label_text = _comparable(hazards[heading.start() :]) if heading else ""
392
+ on_label = {base_code(code) for code in label.hazard_codes}
393
+ # Without the wordings of its own statements: "Giftig für Wasserorganismen,
394
+ # mit langfristiger Wirkung" (H411) stands inside H410's.
395
+ wordings = _catalogue()
396
+ for code in on_label:
397
+ if own := wordings.get(code):
398
+ label_text = label_text.replace(_comparable(own), " ")
399
+ classified = {base_code(code) for code in classification_hazard_codes(hazards)}
400
+ missing: list[Message] = []
401
+ for code in sorted(classification_hazard_codes(hazards)):
402
+ base = base_code(code)
403
+ if base in on_label or _not_required(base, on_label, classified, hazards):
404
+ continue
405
+ # The label prints the statement's wording without its code.
406
+ wording = wordings.get(base)
407
+ if wording and _comparable(wording) in label_text:
408
+ continue
409
+ missing.append(Message("hazard-statement-not-in-label", {"statement": code}))
410
+ return missing
411
+
412
+
256
413
  class Classification(StrEnum):
257
414
  """What a sheet's label states about the product's classification."""
258
415
 
@@ -468,6 +625,7 @@ def sheet_findings(
468
625
  ... "Verordnung (EG) Nr. 1907/2006\\n"
469
626
  ... "Signalwort: Achtung\\n"
470
627
  ... "H315: Verursacht Hautreizungen.\\n"
628
+ ... "P280: Schutzhandschuhe tragen.\\n"
471
629
  ... )
472
630
  >>> result = read_extraction(
473
631
  ... extraction_from_dict(
@@ -540,6 +698,9 @@ def sheet_findings(
540
698
  if contradicts_label(sheet):
541
699
  findings.append(Message("classification-contradicts-label"))
542
700
 
701
+ findings.extend(label_not_found(sheet, parser_warnings))
702
+ findings.extend(hazard_statements_not_in_label(sheet, parser_warnings))
703
+
543
704
  findings.extend(
544
705
  Message("euh-outside-label", dict(outside.params))
545
706
  for outside in euh_outside_label(parser_warnings)
@@ -59,6 +59,7 @@ MESSAGE_CODES: Mapping[str, tuple[str, ...]] = {
59
59
  "no-label-section": (),
60
60
  "several-sheets": ("count",),
61
61
  "classified-before-clp": ("basis",),
62
+ "label-before-clp": (),
62
63
  "no-hazard-codes": (),
63
64
  "label-undetermined": (),
64
65
  "no-precautionary-codes": (),
@@ -84,6 +85,10 @@ MESSAGE_CODES: Mapping[str, tuple[str, ...]] = {
84
85
  "old-format": ("revised",),
85
86
  "classification-not-stated": (),
86
87
  "classification-contradicts-label": (),
88
+ "label-not-found": (),
89
+ "label-without-hazard-statement": (),
90
+ "hazard-statement-not-in-label": ("statement",),
91
+ "label-without-precautionary-statement": (),
87
92
  "different-products": ("products",),
88
93
  "country-editions-differ": ("fields",),
89
94
  "euh-outside-label": ("statement", "sentence"),
@@ -41,6 +41,7 @@ date = "%d.%m.%Y"
41
41
  "line-too-long" = "{count} Zeilen mit mehr als 2.000 Zeichen wurden nicht gelesen, die längste mit {length} Zeichen; ein Blatt druckt keine so langen Zeilen."
42
42
  "no-label-section" = "Abschnitt 2.2 (Kennzeichnungselemente) wurde nicht gefunden; stattdessen wurde der ganze Abschnitt 2 gelesen."
43
43
  "several-sheets" = "Die Datei enthält {count} Sicherheitsdatenblätter; ihre Kennzeichnungen sind zusammengeführt."
44
+ "label-before-clp" = "Die Kennzeichnung (2.2) folgt den Richtlinien vor der CLP-Verordnung: ohne H-, P- und EUH-Code, mit R- oder S-Sätzen, einem Gefahrensymbol oder dem Verweis auf die EG-Richtlinien."
44
45
  "classified-before-clp" = "Abschnitt 2 stuft nach den Richtlinien vor der CLP-Verordnung ein ({basis}); es wurde keine Kennzeichnung gelesen."
45
46
  "no-hazard-codes" = "Die Kennzeichnung trägt keinen H-Satz."
46
47
  "label-undetermined" = "Die Kennzeichnung sagt, die Einstufung sei nicht bestimmt; das Blatt nennt für das Produkt keine Kennzeichnung."
@@ -67,6 +68,10 @@ date = "%d.%m.%Y"
67
68
  "old-format" = "Überarbeitet am {revised}, aber aufgebaut wie vor der Verordnung (EU) 2020/878: keine endokrinschädlichen Eigenschaften, kein Unterabschnitt 11.2, kein Unterabschnitt 12.7."
68
69
  "classification-not-stated" = "Abschnitt 2 nennt weder eine Gefahrenklasse noch sagt er, das Produkt sei nicht eingestuft, und die Kennzeichnung trägt weder H-Satz noch Signalwort noch Piktogramm."
69
70
  "classification-contradicts-label" = "Abschnitt 2 nennt das Produkt nicht eingestuft, und die Kennzeichnung trägt einen H-Satz, ein Signalwort oder ein Piktogramm: Das Blatt widerspricht sich, oder eine der beiden Stellen ist falsch gelesen."
71
+ "label-not-found" = "Abschnitt 2 stuft das Produkt ein, und eine Kennzeichnung (2.2) wurde nicht gefunden; Anhang II der REACH-Verordnung verlangt sie dort."
72
+ "label-without-hazard-statement" = "Abschnitt 2 stuft das Produkt ein, und in der Kennzeichnung wurde kein H-Satz gefunden; die Einstufung verlangt ihre Gefahrenhinweise auf dem Etikett (CLP Art. 27)."
73
+ "label-without-precautionary-statement" = "Abschnitt 2 stuft das Produkt ein, und in der Kennzeichnung wurde kein P-Satz gefunden; die Einstufung verlangt die einschlägigen Sicherheitshinweise auf dem Etikett (CLP Art. 22)."
74
+ "hazard-statement-not-in-label" = "Abschnitt 2.1 nennt {statement} zur Einstufung, und in der Kennzeichnung wurde der Satz nicht gefunden; die Einstufung verlangt ihn auf dem Etikett (CLP Art. 27)."
70
75
  "different-products" = "Die Datei enthält Sicherheitsdatenblätter verschiedener Produkte: {products}"
71
76
  "country-editions-differ" = "Die Blätter desselben Produkts für verschiedene Länder weichen voneinander ab: {fields}"
72
77
  "euh-outside-label" = "{statement} steht nur außerhalb der Kennzeichnung: {sentence}"
@@ -36,6 +36,7 @@
36
36
  "line-too-long" = "{count} lines of more than 2,000 characters were not read, the longest with {length}; a sheet prints no lines that long."
37
37
  "no-label-section" = "Section 2.2 (Kennzeichnungselemente) was not found; the whole of section 2 was read instead."
38
38
  "several-sheets" = "The file holds {count} safety data sheets; their labels are combined."
39
+ "label-before-clp" = "The label (2.2) follows the directives before the CLP Regulation: no H, P or EUH code, but R or S phrases, a hazard symbol or a reference to the EC directives."
39
40
  "classified-before-clp" = "Section 2 classifies under the directives before CLP ({basis}); no label was read."
40
41
  "no-hazard-codes" = "The label carries no H statement."
41
42
  "label-undetermined" = "The label says that the classification was not determined; the sheet states no labelling for the product."
@@ -62,6 +63,10 @@
62
63
  "old-format" = "Revised {revised}, but laid out as before Regulation (EU) 2020/878: no endocrine-disrupting properties, no subsection 11.2, no subsection 12.7."
63
64
  "classification-not-stated" = "Section 2 neither names a hazard class nor says that the product is not classified, and the label carries no H statement, signal word or pictogram."
64
65
  "classification-contradicts-label" = "Section 2 says the product is not classified, and the label carries an H statement, a signal word or a pictogram: the sheet contradicts itself, or one of the two was read wrongly."
66
+ "label-not-found" = "Section 2 classifies the product, and no label (2.2) was found; Annex II of REACH requires it there."
67
+ "label-without-hazard-statement" = "Section 2 classifies the product, and no H statement was found in the label; the classification requires its hazard statements on the label (CLP Article 27)."
68
+ "label-without-precautionary-statement" = "Section 2 classifies the product, and no P statement was found in the label; the classification requires the relevant precautionary statements on the label (CLP Article 22)."
69
+ "hazard-statement-not-in-label" = "Section 2.1 names {statement} for the classification, and the statement was not found in the label; the classification requires it on the label (CLP Article 27)."
65
70
  "different-products" = "The file holds safety data sheets of different products: {products}"
66
71
  "country-editions-differ" = "The sheets of the same product for different countries differ in: {fields}"
67
72
  "euh-outside-label" = "{statement} is printed only outside the label: {sentence}"
@@ -50,6 +50,10 @@ LABEL_UNDETERMINED = "label-undetermined"
50
50
  LABEL_LINE_UNASSIGNED = "label-line-unassigned"
51
51
  EUH_WORDING_OUTSIDE_LABEL = "euh-wording-outside-label"
52
52
  CLASSIFIED_BEFORE_CLP = "classified-before-clp"
53
+ # No block "2.2 Kennzeichnungselemente" was found in section 2 (SDB-4).
54
+ NO_LABEL_SECTION = "no-label-section"
55
+ # The label block follows the directives before CLP (SDB-170).
56
+ LABEL_BEFORE_CLP = "label-before-clp"
53
57
  STATEMENT_TOO_LONG = "statement-too-long"
54
58
  # How many lines a statement of the hint texts spans at most (SDB-168): the
55
59
  # longest a sheet of the measured collection prints has 37.
@@ -179,9 +183,10 @@ SUBSECTION_TITLES: dict[int, re.Pattern[str]] = {
179
183
  # The dot after the number is optional: "3 Zusammensetzung" is as common as
180
184
  # "3. Zusammensetzung", and both are headings. Section numbers stop at 16.
181
185
  # A mark for a changed section may stand in front (SDB-152): "* 1 Bezeichnung
182
- # des …", "! 2. Mögliche Gefahren".
186
+ # des …", "! 2. Mögliche Gefahren". A colon may stand for the dot (SDB-175):
187
+ # "3: Zusammensetzung / Angaben zu den Bestandteilen".
183
188
  NUMBERED_SECTION_PATTERN = re.compile(
184
- r"^(?:[*!]\s*)?0?(1[0-6]|[1-9])\s*\.?\s+(\S.{3,79}?)\s*$"
189
+ r"^(?:[*!]\s*)?0?(1[0-6]|[1-9])\s*[.:]?\s+(\S.{3,79}?)\s*$"
185
190
  )
186
191
  # Suppliers state the document's own version several ways: "Version 8.2",
187
192
  # "Version: 1.1", "Versionsnummer 1", "Nummer der Fassung: 1.0" and
@@ -601,6 +606,32 @@ _HAZARD_DESIGNATION = re.compile(
601
606
  _CLP_CODE_PRINTED = re.compile(r"\b(?:[HP][^\S\n]?\d{3}|EUH[^\S\n]?\d{3})\b")
602
607
 
603
608
 
609
+ # A label under the directives before CLP names them as its basis: "Kennzeichnung
610
+ # nach EWG-Richtlinien", "nach EG-Richtlinien/GefStoffV nicht
611
+ # kennzeichnungspflichtig", "gemäß 1999/45/EG".
612
+ _OLD_RULES_LABEL = re.compile(
613
+ r"\b(?:EWG|EG)-Richtlinien\b|67/548|(?:19)?99/45|\bGefStoffV\b", re.IGNORECASE
614
+ )
615
+
616
+
617
+ def label_before_clp(label: str) -> bool:
618
+ """Whether a label block follows the directives before CLP (SDB-170).
619
+
620
+ It prints no H, P or EUH code, and either names the old directives as its
621
+ basis or labels with R phrases or a hazard symbol as SDB-73 reads them.
622
+
623
+ >>> label_before_clp("Kennzeichnung nach EWG-Richtlinien:\\nnicht kennzeichnungspflichtig.")
624
+ True
625
+ >>> label_before_clp("Kennzeichnung nach EWG-Richtlinien\\nH315 Verursacht Hautreizungen.")
626
+ False
627
+ >>> label_before_clp("Signalwort: Achtung")
628
+ False
629
+ """
630
+ if _CLP_CODE_PRINTED.search(label):
631
+ return False
632
+ return bool(_OLD_RULES_LABEL.search(label)) or old_classification(label) is not None
633
+
634
+
604
635
  def old_classification(hazards_section: str) -> str | None:
605
636
  """What section 2 classifies with under the rules before CLP, else None (SDB-73)."""
606
637
  if _CLP_CODE_PRINTED.search(hazards_section):
@@ -794,6 +825,43 @@ def _gives_no_answer(part: str) -> bool:
794
825
  return found > 0 and not re.search(r"[^\W\d]", rest)
795
826
 
796
827
 
828
+ # A hazard statement's code as section 2.1 names it with a hazard class: H200
829
+ # to H499, H360 and H361 with the letters of their effect (SDB-173).
830
+ _CLASSIFICATION_HAZARD_CODE = re.compile(
831
+ r"\bH(?:36[01][A-Za-z]{1,2}(?![^\W\d_])|[2-4]\d{2})(?!\d)"
832
+ )
833
+
834
+
835
+ def classification_hazard_codes(hazards_section: str) -> set[str]:
836
+ r"""The H codes section 2.1 names for the product's classification (SDB-173).
837
+
838
+ Read from the text under each classification heading up to the label's
839
+ (SDB-75); never codes of the sheet. A code on a line that speaks of
840
+ ingredients ("enthält", "Inhaltsstoff", "Bestandteil", as in SDB-80), or on
841
+ the line after one that it continues, is an ingredient's, not the product's.
842
+
843
+ >>> classification_hazard_codes(
844
+ ... "2.1 Einstufung des Stoffs oder Gemischs\n"
845
+ ... "Flam. Liq. 3, H226\nRepr. 1B, H360D\n"
846
+ ... "Das Gemisch enthält Stoffe, die als Acute Tox. 4\n(H302) eingestuft sind.\n"
847
+ ... "2.2 Kennzeichnungselemente\nH226 Flüssigkeit und Dampf entzündbar.\n"
848
+ ... ) == {"H226", "H360D"}
849
+ True
850
+ """
851
+ codes: set[str] = set()
852
+ for part in _classification_parts(hazards_section):
853
+ lines = part.split("\n")
854
+ for index, line in enumerate(lines):
855
+ found = _CLASSIFICATION_HAZARD_CODE.findall(line)
856
+ if not found or _ABOUT_INGREDIENTS.search(line):
857
+ continue
858
+ continues = line.lstrip()[:1].islower() or line.lstrip().startswith("(")
859
+ if continues and index and _ABOUT_INGREDIENTS.search(lines[index - 1]):
860
+ continue
861
+ codes.update(lettered_code(code) for code in found)
862
+ return codes
863
+
864
+
797
865
  def stated_classification(hazards_section: str) -> StatedClassification:
798
866
  """What section 2 says about the product's classification (SDB-75).
799
867
 
@@ -1341,8 +1409,9 @@ class SDSParser:
1341
1409
  def read_numbered_section_heading(line: str) -> int | None:
1342
1410
  """Return the section number of a plainly numbered heading, else None.
1343
1411
 
1344
- Recognises "2. Mögliche Gefahren" and "2 Mögliche Gefahren", the two forms
1345
- that carry no "Abschnitt" for SECTION_HEADER_PATTERN to find. The title must
1412
+ Recognises "2. Mögliche Gefahren", "2: Mögliche Gefahren" and "2 Mögliche
1413
+ Gefahren", the forms that carry no "Abschnitt" for SECTION_HEADER_PATTERN
1414
+ to find. The title must
1346
1415
  agree with the number: see SECTION_TITLE_PATTERNS for why the number alone
1347
1416
  is not enough.
1348
1417
  """
@@ -1979,9 +2048,11 @@ def _parse_hazard_statements(
1979
2048
  ]
1980
2049
  labels = [label for _, label in labelled]
1981
2050
  if not labels and full_text:
1982
- warnings.append(Message("no-label-section"))
2051
+ warnings.append(Message(NO_LABEL_SECTION))
1983
2052
  if len(labels) > 1:
1984
2053
  warnings.append(Message("several-sheets", {"count": len(labels)}))
2054
+ if labels and all(label_before_clp(label) for label in labels):
2055
+ warnings.append(Message(LABEL_BEFORE_CLP))
1985
2056
 
1986
2057
  # Each sheet's label is read as if it stood alone, then the product carries
1987
2058
  # all of them. A combination pack holds both products, so it carries both
@@ -2045,6 +2116,18 @@ def _parse_hazard_statements(
2045
2116
  signal_word=_strongest_signal_word(r.signal_word for r in readings),
2046
2117
  pictograms=sorted(set().union(*(r.pictograms for r in readings))),
2047
2118
  )
2119
+ # A code one sheet's label prints is printed, even where another sheet of
2120
+ # the file only prints its sentence (SDB-174).
2121
+ printed_somewhere = set().union(
2122
+ *(
2123
+ (r.hazard_codes | r.precautionary_codes | r.supplemental) - r.from_wording
2124
+ for r in readings
2125
+ )
2126
+ )
2127
+ from_wording = (
2128
+ set().union(*(r.from_wording for r in readings))
2129
+ | {code for code, _ in worded_supplemental}
2130
+ ) - printed_somewhere
2048
2131
  # The text has to say what the codes say. The codes are read off the label;
2049
2132
  # the text is cut from the whole hazards section, which also holds 2.1's
2050
2133
  # classification and, where a heading goes unread, the section 3 ingredient
@@ -2119,6 +2202,12 @@ def _parse_hazard_statements(
2119
2202
  supplemental.add(code.group(0).upper())
2120
2203
  supplemental_text = "\n".join(kept) or None
2121
2204
  hazard_statements.supplemental_codes = sorted(supplemental)
2205
+ # A code kept by its wording here is printed in the text (SDB-124).
2206
+ from_wording -= {
2207
+ code.group(0).upper()
2208
+ for line in kept
2209
+ if (code := SUPPLEMENTAL_CODE_PATTERN.match(line)) is not None
2210
+ }
2122
2211
  if supplemental_text:
2123
2212
  supplemental_text = _with_substances_before_the_code(
2124
2213
  supplemental_text, sections.get(Section.HAZARDS, "")
@@ -2139,6 +2228,12 @@ def _parse_hazard_statements(
2139
2228
  )
2140
2229
  or None
2141
2230
  )
2231
+ on_label = {
2232
+ *hazard_statements.hazard_codes,
2233
+ *hazard_statements.precautionary_codes,
2234
+ *hazard_statements.supplemental_codes,
2235
+ }
2236
+ hazard_statements.codes_from_wording = sorted(from_wording & on_label)
2142
2237
  _without_trailing_pictograms(hazard_statements)
2143
2238
  warnings.extend(Message(STATEMENT_TOO_LONG, {"codes": codes}) for codes in too_long)
2144
2239
  # What the label holds that nothing above read (SDB-165): a measure of
@@ -2514,6 +2609,9 @@ class _LabelReading:
2514
2609
  # The sheet's sentence read as P501 by SDB-166, where the label prints no
2515
2610
  # P501 code.
2516
2611
  disposal: str | None = None
2612
+ # The codes the parser assigned to sentences the label prints without
2613
+ # them (SDB-174): SDB-6, SDB-28, SDB-163, SDB-166.
2614
+ from_wording: set[str] = field(default_factory=set)
2517
2615
 
2518
2616
 
2519
2617
  def _strongest_signal_word(words: Iterable[str | None]) -> str | None:
@@ -2593,6 +2691,9 @@ def _read_label(
2593
2691
  hazard_section_text = _LABEL_DECLINED.sub("", hazard_section_text)
2594
2692
 
2595
2693
  hazard_codes = sorted(printed_hazard_codes(hazard_section_text))
2694
+ # Which list came from the sentences rather than printed codes (SDB-174).
2695
+ hazard_by_wording = False
2696
+ precautionary_by_wording: set[str] = set()
2596
2697
  if not hazard_codes and classification is not None:
2597
2698
  hazard_codes = sorted(
2598
2699
  {lettered_code(code) for code in LABEL_HAZARD_CODE.findall(classification)}
@@ -2623,6 +2724,7 @@ def _read_label(
2623
2724
  # "H", not "not EUH": EUH statements have their own field, and P
2624
2725
  # statements are somebody else's list entirely.
2625
2726
  hazard_codes = sorted(c for c in by_wording if c.startswith("H"))
2727
+ hazard_by_wording = bool(hazard_codes)
2626
2728
  if not precautionary_codes:
2627
2729
  by_variants: set[str] = set()
2628
2730
  # And in the forms Annex IV allows for its choices, details and
@@ -2639,6 +2741,7 @@ def _read_label(
2639
2741
  precautionary_codes = sorted(
2640
2742
  {c for c in by_wording if c.startswith("P")} | by_variants
2641
2743
  )
2744
+ precautionary_by_wording = set(precautionary_codes)
2642
2745
  # P501 by the words many suppliers use for it, also beside other P codes
2643
2746
  # (SDB-166); never on a label by the rules before CLP (SDB-120).
2644
2747
  disposal = None
@@ -2653,6 +2756,7 @@ def _read_label(
2653
2756
  disposal = _disposal_sentence(hazard_section_text)
2654
2757
  if disposal is not None:
2655
2758
  precautionary_codes = sorted({*precautionary_codes, "P501"})
2759
+ precautionary_by_wording.add("P501")
2656
2760
  supplemental = sorted(
2657
2761
  set(
2658
2762
  code.upper()
@@ -2713,6 +2817,9 @@ def _read_label(
2713
2817
  pictograms=set(pictograms),
2714
2818
  warnings=warnings,
2715
2819
  disposal=disposal,
2820
+ from_wording=(h_code_set if hazard_by_wording else set())
2821
+ | precautionary_by_wording
2822
+ | {code for code, _ in worded_supplemental},
2716
2823
  )
2717
2824
 
2718
2825
 
@@ -362,6 +362,13 @@
362
362
  "string",
363
363
  "null"
364
364
  ]
365
+ },
366
+ "codes_from_wording": {
367
+ "type": "array",
368
+ "items": {
369
+ "type": "string"
370
+ },
371
+ "description": "The codes the label does not print and the parser assigned to the sentences it prints (SDB-174); each also stands in its list. Missing in a reading stored before."
365
372
  }
366
373
  }
367
374
  },
@@ -460,6 +467,14 @@
460
467
  "null"
461
468
  ],
462
469
  "description": "The wording of the CLP Regulation, where it has exactly one."
470
+ },
471
+ "code_source": {
472
+ "type": "string",
473
+ "enum": [
474
+ "printed",
475
+ "from_wording"
476
+ ],
477
+ "description": "Whether the label prints the code, or prints only the sentence and the parser assigned the code (SDB-174)."
463
478
  }
464
479
  }
465
480
  },
@@ -509,6 +524,10 @@
509
524
  "old-format",
510
525
  "classification-not-stated",
511
526
  "classification-contradicts-label",
527
+ "label-not-found",
528
+ "label-without-hazard-statement",
529
+ "label-without-precautionary-statement",
530
+ "hazard-statement-not-in-label",
512
531
  "different-products",
513
532
  "country-editions-differ",
514
533
  "euh-outside-label",
@@ -683,9 +702,14 @@
683
702
  "embedded-wrong-root",
684
703
  "euh-outside-label",
685
704
  "euh-wording-outside-label",
705
+ "label-before-clp",
686
706
  "label-line-unassigned",
707
+ "label-not-found",
687
708
  "statement-too-long",
688
709
  "label-undetermined",
710
+ "label-without-hazard-statement",
711
+ "label-without-precautionary-statement",
712
+ "hazard-statement-not-in-label",
689
713
  "leaflet-title",
690
714
  "line-too-long",
691
715
  "model-answer-unreadable",
@@ -41,6 +41,9 @@ class HazardStatements:
41
41
  hazard_text: str | None = None
42
42
  precautionary_text: str | None = None
43
43
  supplemental_text: str | None = None
44
+ # The codes the label does not print and the parser assigned to the
45
+ # sentences it prints (SDB-174); every one of them also stands in its list.
46
+ codes_from_wording: list[str] = field(default_factory=list)
44
47
 
45
48
 
46
49
  @dataclass(slots=True)
@@ -90,6 +90,7 @@ def _sheet_fields(sheet: SafetyDataSheet, content_hash: str | None) -> dict[str,
90
90
  "hazard_text": sheet.hazard_statements.hazard_text,
91
91
  "precautionary_text": sheet.hazard_statements.precautionary_text,
92
92
  "supplemental_text": sheet.hazard_statements.supplemental_text,
93
+ "codes_from_wording": sheet.hazard_statements.codes_from_wording,
93
94
  },
94
95
  "contact_info": {
95
96
  "supplier_name": sheet.contact_info.supplier_name,
@@ -163,6 +164,8 @@ def sheet_from_dict(stored: Mapping[str, Any]) -> SafetyDataSheet:
163
164
  hazard_text=_optional_text(label["hazard_text"]),
164
165
  precautionary_text=_optional_text(label["precautionary_text"]),
165
166
  supplemental_text=_optional_text(label["supplemental_text"]),
167
+ # Not in a reading stored before SDB-174.
168
+ codes_from_wording=_texts(label.get("codes_from_wording", [])),
166
169
  ),
167
170
  contact_info=ContactInfo(
168
171
  supplier_name=_optional_text(contact["supplier_name"]),
@@ -308,7 +311,12 @@ def reading_to_dict(
308
311
  def _statements(sheet: SafetyDataSheet) -> list[dict[str, Any]]:
309
312
  """The label's statements, each with both wordings (SDW-5)."""
310
313
  return [
311
- {"code": s.code, "sheet": s.sheet, "regulation": s.regulation}
314
+ {
315
+ "code": s.code,
316
+ "sheet": s.sheet,
317
+ "regulation": s.regulation,
318
+ "code_source": s.code_source,
319
+ }
312
320
  for s in label_statements(sheet.hazard_statements)
313
321
  ]
314
322
 
@@ -1041,6 +1041,50 @@ class StatementWording:
1041
1041
  sheet: str | None
1042
1042
  # None where the regulation gives the code no single wording.
1043
1043
  regulation: str | None
1044
+ # PRINTED where the label prints the code, FROM_WORDING where it prints
1045
+ # only the sentence and the parser assigned the code (SDB-174).
1046
+ code_source: str = "printed"
1047
+
1048
+
1049
+ PRINTED = "printed"
1050
+ FROM_WORDING = "from_wording"
1051
+
1052
+
1053
+ def _assigned(code: str, from_wording: Collection[str]) -> bool:
1054
+ """Whether every code of *code* ("P305+P351+P338" counts as its parts)
1055
+ is one the parser assigned to a sentence (SDB-174)."""
1056
+ return all(part in from_wording for part in code.split("+"))
1057
+
1058
+
1059
+ def as_printed(text: str | None, codes_from_wording: Collection[str]) -> str | None:
1060
+ r"""*text* with the codes the label does not print taken off its lines (SDB-174).
1061
+
1062
+ A line whose leading code -- every part of a combination -- the parser
1063
+ assigned to the sheet's sentence keeps the sentence and loses the code;
1064
+ every other line stays as it is.
1065
+
1066
+ >>> as_printed(
1067
+ ... "P280: Schutzhandschuhe tragen.\nP501 Inhalt/Behälter … entsorgen",
1068
+ ... ["P501"],
1069
+ ... )
1070
+ 'P280: Schutzhandschuhe tragen.\nInhalt/Behälter … entsorgen'
1071
+ >>> as_printed("EUH208: Enthält Musterol. Kann allergische Reaktionen "
1072
+ ... "hervorrufen.", ["EUH208"])
1073
+ 'Enthält Musterol. Kann allergische Reaktionen hervorrufen.'
1074
+ >>> as_printed("P305+P351+P338: BEI KONTAKT MIT DEN AUGEN: …", ["P338"])
1075
+ 'P305+P351+P338: BEI KONTAKT MIT DEN AUGEN: …'
1076
+ """
1077
+ if not text or not codes_from_wording:
1078
+ return text
1079
+ lines = []
1080
+ for line in text.split("\n"):
1081
+ statement = STATEMENT_LINE.match(line)
1082
+ if statement is not None and statement.group(4).strip():
1083
+ code = re.sub(r"\s*\+\s*", "+", statement.group(2))
1084
+ if _assigned(code.upper(), codes_from_wording):
1085
+ line = statement.group(1) + statement.group(4).strip()
1086
+ lines.append(line)
1087
+ return "\n".join(lines)
1044
1088
 
1045
1089
 
1046
1090
  # Codes standing first in a wording: the rest of a row of codes joined by
@@ -1112,8 +1156,14 @@ def label_statements(hazards: HazardStatements) -> list[StatementWording]:
1112
1156
 
1113
1157
  for code in sorted(on_label - {part for c in wordings for part in c.split("+")}):
1114
1158
  add(code, None)
1159
+ assigned = set(hazards.codes_from_wording)
1115
1160
  return [
1116
- StatementWording(code, wording, canonical_text(code))
1161
+ StatementWording(
1162
+ code,
1163
+ wording,
1164
+ canonical_text(code),
1165
+ FROM_WORDING if assigned and _assigned(code, assigned) else PRINTED,
1166
+ )
1117
1167
  for code, printed in wordings.items()
1118
1168
  for wording in (printed or (None,))
1119
1169
  ]
@@ -19,7 +19,7 @@ EXTRACTION_VERSION = 2
19
19
  # Raise when the same text can give another reading: the verdict, the parser,
20
20
  # the statements and their tables, the findings, the message codes, the JSON
21
21
  # form. The sentences of the catalogs are no part of a reading.
22
- PARSER_VERSION = 19
22
+ PARSER_VERSION = 20
23
23
 
24
24
 
25
25
  def extraction_stage() -> str:
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.4
2
2
  Name: sdb-reader
3
- Version: 0.10.0
3
+ Version: 0.11.0
4
4
  Summary: Reads German safety data sheets (Sicherheitsdatenblätter) under REACH and CLP, without a language model
5
5
  Author-email: "Förde-Baumarkt Holger Claussen GmbH & Co. KG" <service@foerde-baumarkt.de>
6
6
  License-Expression: AGPL-3.0-or-later
File without changes
File without changes
File without changes
File without changes
File without changes