20120615911

Committed 10 Dec 2025 10:24PM UTC coverage: 89.767% (+0.08%) from 89.689%

Build # 20120615911

Build Type

push

github

Committed by

bramp

Commit Message

Fix typo in fixtures README: .sh -> .py for regenerate script

Coverage Stats

11790 of 13134 relevant lines covered (89.77%)

0.9 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

96.67

/src/build_a_long/pdf_extract/classifier/steps/step_count_classifier.py

"""
Step count classifier.

Purpose
-------
Detect step-count text like "2x" that appears in substep callout boxes.
These are similar to part counts but use a larger font size (typically 16pt),
between part count size and step number size.

Debugging
---------
Enable DEBUG logs with LOG_LEVEL=DEBUG.
"""

import logging
from typing import ClassVar

from build_a_long.pdf_extract.classifier.candidate import Candidate
from build_a_long.pdf_extract.classifier.classification_result import (
    ClassificationResult,
)
from build_a_long.pdf_extract.classifier.rule_based_classifier import (
    RuleBasedClassifier,
)
from build_a_long.pdf_extract.classifier.rules import (
    FontSizeRangeRule,
    IsInstanceFilter,
    PartCountTextRule,
    Rule,
)
from build_a_long.pdf_extract.classifier.text import (
    extract_part_count_value,
)
from build_a_long.pdf_extract.extractor.lego_page_elements import (
    StepCount,
)
from build_a_long.pdf_extract.extractor.page_blocks import Text

log = logging.getLogger(__name__)


class StepCountClassifier(RuleBasedClassifier):
    """Classifier for step counts (substep counts like "2x").

    These are count labels that appear inside substep callout boxes,
    indicating how many times to build the sub-assembly.
    They use a font size between part counts and step numbers.
    """

    output: ClassVar[str] = "step_count"
    requires: ClassVar[frozenset[str]] = frozenset()

    @property
    def min_score(self) -> float:
        return self.config.step_count.min_score

    @property
    def rules(self) -> list[Rule]:
        config = self.config
        step_count_config = config.step_count
        hints = config.font_size_hints

        return [
            # Must be text
            IsInstanceFilter(Text),
            # Check if text matches count pattern (e.g., "2x", "4x")
            PartCountTextRule(
                weight=step_count_config.text_weight,
                name="text_score",
                required=True,
            ),
            # Score font size: should be >= part_count_size and <= step_number_size
            FontSizeRangeRule(
                min_size=hints.part_count_size,
                max_size=hints.step_number_size,
                tolerance=1.0,
                weight=step_count_config.font_size_weight,
                name="font_size_score",
            ),
        ]

    def build(self, candidate: Candidate, result: ClassificationResult) -> StepCount:
        """Construct a StepCount element from a candidate.

        The candidate may include additional source blocks (e.g., text outline
        effects) beyond the primary Text block.
        """
        # Get the primary text block (first in source_blocks)
        assert len(candidate.source_blocks) >= 1
        block = candidate.source_blocks[0]
        assert isinstance(block, Text)

        # Parse the count value
        value = extract_part_count_value(block.text)
        if value is None:
            raise ValueError(f"Could not parse step count from text: '{block.text}'")

        return StepCount(count=value, bbox=block.bbox)

1	"""
2	Step count classifier.
3
4	Purpose
5	-------
6	Detect step-count text like "2x" that appears in substep callout boxes.
7	These are similar to part counts but use a larger font size (typically 16pt),
8	between part count size and step number size.
9
10	Debugging
11	---------
12	Enable DEBUG logs with LOG_LEVEL=DEBUG.
13	"""
14
15	import logging	1✔
16	from typing import ClassVar	1✔
17
18	from build_a_long.pdf_extract.classifier.candidate import Candidate	1✔
19	from build_a_long.pdf_extract.classifier.classification_result import (	1✔
20	ClassificationResult,
21	)
22	from build_a_long.pdf_extract.classifier.rule_based_classifier import (	1✔
23	RuleBasedClassifier,
24	)
25	from build_a_long.pdf_extract.classifier.rules import (	1✔
26	FontSizeRangeRule,
27	IsInstanceFilter,
28	PartCountTextRule,
29	Rule,
30	)
31	from build_a_long.pdf_extract.classifier.text import (	1✔
32	extract_part_count_value,
33	)
34	from build_a_long.pdf_extract.extractor.lego_page_elements import (	1✔
35	StepCount,
36	)
37	from build_a_long.pdf_extract.extractor.page_blocks import Text	1✔
38
39	log = logging.getLogger(__name__)	1✔
40
41
42	class StepCountClassifier(RuleBasedClassifier):	1✔
43	"""Classifier for step counts (substep counts like "2x").
44
45	These are count labels that appear inside substep callout boxes,
46	indicating how many times to build the sub-assembly.
47	They use a font size between part counts and step numbers.
48	"""
49
50	output: ClassVar[str] = "step_count"	1✔
51	requires: ClassVar[frozenset[str]] = frozenset()	1✔
52
53	@property	1✔
54	def min_score(self) -> float:	1✔
55	return self.config.step_count.min_score	1✔
56
57	@property	1✔
58	def rules(self) -> list[Rule]:	1✔
59	config = self.config	1✔
60	step_count_config = config.step_count	1✔
61	hints = config.font_size_hints	1✔
62
63	return [	1✔
64	# Must be text
65	IsInstanceFilter(Text),
66	# Check if text matches count pattern (e.g., "2x", "4x")
67	PartCountTextRule(
68	weight=step_count_config.text_weight,
69	name="text_score",
70	required=True,
71	),
72	# Score font size: should be >= part_count_size and <= step_number_size
73	FontSizeRangeRule(
74	min_size=hints.part_count_size,
75	max_size=hints.step_number_size,
76	tolerance=1.0,
77	weight=step_count_config.font_size_weight,
78	name="font_size_score",
79	),
80	]
81
82	def build(self, candidate: Candidate, result: ClassificationResult) -> StepCount:	1✔
83	"""Construct a StepCount element from a candidate.
84
85	The candidate may include additional source blocks (e.g., text outline
86	effects) beyond the primary Text block.
87	"""
88	# Get the primary text block (first in source_blocks)
89	assert len(candidate.source_blocks) >= 1	1✔
90	block = candidate.source_blocks[0]	1✔
91	assert isinstance(block, Text)	1✔
92
93	# Parse the count value
94	value = extract_part_count_value(block.text)	1✔
95	if value is None:	1✔
96	raise ValueError(f"Could not parse step count from text: '{block.text}'")	×
97
98	return StepCount(count=value, bbox=block.bbox)	1✔

bramp / build-along / 20120615911

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous