22072536647

Committed 16 Feb 2026 05:41PM UTC coverage: 42.093%. First build

Build # 22072536647

Build Type

Pull #59

github

Pull Request Pull Request #59: apps/summerization: Integrate Document Summary into Workflow

Coverage Stats

51 of 314 new or added lines in 7 files covered. (16.24%)

3564 of 8467 relevant lines covered (42.09%)

0.42 hits per line

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

88.46

/apps/summarization/pydantic_models.py

"""Pydantic models for summarization responses."""

from typing import List
from typing import Literal
from typing import Optional

from pydantic import BaseModel
from pydantic import Field


class SummaryItem(BaseModel):
    """Response model for summarization."""

    title: str = Field(description="Title of the summary")
    summary: str = Field(description="The summary of the text or document")
    key_points: list[str] = Field(
        default_factory=list,
        description="Important points or keywords extracted from the text or document",
    )


class ModuleItem(BaseModel):
    """Response model for module summarization."""

    module_name: str = Field(description="Name of the module")
    summary: str = Field(description="Summary of the module")
    key_points: list[str] = Field(
        default_factory=list, description="Key points of the module"
    )
    phase_status: str = Field(
        description="Phase status: 'past' (in the past), 'active' (currently running), 'upcoming' (in the future)"
    )
    link: str = Field(description="Link to the module")


class SummaryResponse(BaseModel):
    """Response model for summarization."""

    summary_items: list[SummaryItem] = Field(
        default_factory=list,
        description=(
            "List of summary items. Each item contains: "
            "title, summary, key_points (list of important points)"
        ),
    )
    module_items: list[ModuleItem] = Field(
        default_factory=list,
        description=(
            "List of module items. Each item contains: "
            "module_name, summary, key_points (list of important points), "
            "phase_status (past/active/upcoming), link (URL to the module)"
        ),
    )


"""Pydantic models for summarization responses."""


# Debug sub-models
class Claim(BaseModel):
    """A claim extracted from the summary with evidence."""

    claim_text: str = Field(description="The claim text")
    evidence_type: Literal[
        "from_votes",
        "from_ratings",
        "from_open_answers",
        "from_comments",
        "from_base_text",
        "uncertain",
    ] = Field(description="Type of evidence supporting this claim")
    action: Literal["keep", "soften", "replace", "remove"] = Field(
        description="Action taken on this claim"
    )
    fix_hint: Optional[str] = Field(None, description="Hint for fixing the claim")


class QuantifierFix(BaseModel):
    """A fix for an uncertain quantifier."""

    original_phrase: str = Field(description="Original uncertain phrase")
    replacement: str = Field(description="Replacement phrase")
    reason: str = Field(description="Reason for the fix")


class Patch(BaseModel):
    """A patch applied to the summary."""

    patch_type: Literal["REPLACE", "REMOVE", "ADD_SENTENCE"] = Field(
        description="Type of patch applied"
    )
    target: str = Field(description="Target text to patch")
    replacement: Optional[str] = Field(
        None, description="Replacement text (for REPLACE/ADD)"
    )


class ModuleDebug(BaseModel):
    """Debug information for a module summary."""

    module_type: str = Field(description="Type of module")
    signals_snapshot: List[str] = Field(
        default_factory=list, description="Snapshot of signals at generation time"
    )
    draft_before_qa: str = Field(description="Draft summary before QA")
    claims: List[Claim] = Field(
        default_factory=list, description="Claims extracted from summary"
    )
    quantifier_fixes: List[QuantifierFix] = Field(
        default_factory=list, description="Fixes for uncertain quantifiers"
    )
    anchors: List[str] = Field(
        default_factory=list, description="Anchor points in the data"
    )
    coverage_gaps: List[str] = Field(
        default_factory=list, description="Identified gaps in coverage"
    )
    coverage_patch: Optional[str] = Field(
        None, description="Patch to fix coverage gaps"
    )
    patches: List[Patch] = Field(default_factory=list, description="Patches applied")
    after_qa: str = Field(description="Summary after QA process")
    diff_summary: Optional[str] = Field(None, description="Summary of changes made")
    qa_status: Literal["PASS", "FAIL"] = Field(description="QA status")


class ModuleFinal(BaseModel):
    """Final summary and bullets for a module."""

    summary: str = Field(description="Summary of the module's outcomes/activities")
    bullets: List[str] = Field(
        default_factory=list,
        description="Key points or main sentiments from the module",
    )


class PhaseModule(BaseModel):
    """Module within a phase section."""

    module_name: str = Field(description="Name of the module")
    module_id: int = Field(description="ID of the module")
    status: Literal["past", "current", "upcoming"] = Field(description="Module status")
    debug: Optional[ModuleDebug] = Field(None, description="Debug information")
    final: ModuleFinal = Field(description="Summary and key points for the module")


class PhaseSection(BaseModel):
    """A phase section containing modules."""

    modules: List[PhaseModule] = Field(
        default_factory=list, description="Modules in this phase"
    )


class Phases(BaseModel):
    """All phase sections."""

    past: PhaseSection = Field(default_factory=PhaseSection)
    current: PhaseSection = Field(default_factory=PhaseSection)
    upcoming: PhaseSection = Field(default_factory=PhaseSection)


class GeneralInfo(BaseModel):
    """General information about the project."""

    summary: str = Field(description="General summary of the entire project")
    goals: List[str] = Field(
        default_factory=list, description="Overall goals of the project"
    )


class ProjectSummaryResponse(BaseModel):
    """Response model for complete project summarization."""

    title: str = Field(default="Summary of participation")
    general_info: GeneralInfo = Field(description="General project information")
    phases: Phases = Field(description="All phase sections with their modules")


class DocumentInputItem(BaseModel):
    """Single document input item with handle and URL."""

    handle: str = Field(description="Unique identifier/handle for the document")
    url: str = Field(description="URL of the document")

    def is_image(self) -> bool:
        """Check if the URL points to an image file."""
        image_extensions = (
            ".jpg",
            ".jpeg",
            ".png",
            ".gif",
            ".webp",
            ".mpo",
            ".heif",
            ".avif",
            ".bmp",
            ".tiff",
            ".tif",
        )
        url_lower = self.url.lower()
        return any(url_lower.endswith(ext) for ext in image_extensions)

    def is_document(self) -> bool:
        """Check if the URL points to a document file (non-image)."""
        document_extensions = (".pdf", ".doc", ".docx", ".txt", ".rtf", ".odt")
        url_lower = self.url.lower()
        return any(url_lower.endswith(ext) for ext in document_extensions)


class DocumentSummaryItem(BaseModel):
    """Response model for a single document summary with handle."""

    handle: str = Field(description="Unique identifier/handle for the document")
    summary: str = Field(description="Summary of the document content")


class DocumentSummaryResponse(BaseModel):
    """Response model for multiple document summaries."""

    documents: list[DocumentSummaryItem] = Field(
        default_factory=list,
        description="List of document summaries, each with handle and summary",
    )

1	"""Pydantic models for summarization responses."""
2
3	from typing import List	1✔
4	from typing import Literal
5	from typing import Optional	1✔
6		1✔
7	from pydantic import BaseModel
8	from pydantic import Field
9		1✔
10
11	class SummaryItem(BaseModel):
12	"""Response model for summarization."""	1✔
13		1✔
14	title: str = Field(description="Title of the summary")	1✔
15	summary: str = Field(description="The summary of the text or document")
16	key_points: list[str] = Field(
17	default_factory=list,
18	description="Important points or keywords extracted from the text or document",
19	)
20		1✔
21
22	class ModuleItem(BaseModel):
23	"""Response model for module summarization."""	1✔
24		1✔
25	module_name: str = Field(description="Name of the module")	1✔
26	summary: str = Field(description="Summary of the module")
27	key_points: list[str] = Field(
28	default_factory=list, description="Key points of the module"	1✔
29	)
30	phase_status: str = Field(
31	description="Phase status: 'past' (in the past), 'active' (currently running), 'upcoming' (in the future)"	1✔
32	)
33	link: str = Field(description="Link to the module")
34		1✔
35
36	class SummaryResponse(BaseModel):
37	"""Response model for summarization."""	1✔
38
39	summary_items: list[SummaryItem] = Field(
40	default_factory=list,
41	description=(
42	"List of summary items. Each item contains: "
43	"title, summary, key_points (list of important points)"
44	),	1✔
45	)
46	module_items: list[ModuleItem] = Field(
47	default_factory=list,
48	description=(
49	"List of module items. Each item contains: "
50	"module_name, summary, key_points (list of important points), "
51	"phase_status (past/active/upcoming), link (URL to the module)"
52	),
53	)
54		1✔
55
56	"""Pydantic models for summarization responses."""
57		1✔
58
59	# Debug sub-models
60	class Claim(BaseModel):	1✔
61	"""A claim extracted from the summary with evidence."""	1✔
62		1✔
63	claim_text: str = Field(description="The claim text")
64	evidence_type: Literal[
65	"from_votes",	1✔
66	"from_ratings",
67	"from_open_answers",
68	"from_comments",	1✔
69	"from_base_text",	1✔
70	"uncertain",	1✔
71	] = Field(description="Type of evidence supporting this claim")
72	action: Literal["keep", "soften", "replace", "remove"] = Field(
73	description="Action taken on this claim"
74	)	1✔
75	fix_hint: Optional[str] = Field(None, description="Hint for fixing the claim")
76
77		1✔
78	class QuantifierFix(BaseModel):	1✔
79	"""A fix for an uncertain quantifier."""
80
81	original_phrase: str = Field(description="Original uncertain phrase")
82	replacement: str = Field(description="Replacement phrase")
83	reason: str = Field(description="Reason for the fix")	1✔
84
85
86	class Patch(BaseModel):
87	"""A patch applied to the summary."""	1✔
88
89	patch_type: Literal["REPLACE", "REMOVE", "ADD_SENTENCE"] = Field(
90	description="Type of patch applied"	1✔
91	)
92	target: str = Field(description="Target text to patch")	1✔
93	replacement: Optional[str] = Field(	1✔
94	None, description="Replacement text (for REPLACE/ADD)"
95	)	1✔
96
97
98	class ModuleDebug(BaseModel):
99	"""Debug information for a module summary."""
100		1✔
101	module_type: str = Field(description="Type of module")
102	signals_snapshot: List[str] = Field(
103	default_factory=list, description="Snapshot of signals at generation time"
104	)
105	draft_before_qa: str = Field(description="Draft summary before QA")	1✔
106	claims: List[Claim] = Field(
107	default_factory=list, description="Claims extracted from summary"
108	)
109	quantifier_fixes: List[QuantifierFix] = Field(
110	default_factory=list, description="Fixes for uncertain quantifiers"
111	)	1✔
112	anchors: List[str] = Field(
113	default_factory=list, description="Anchor points in the data"
114	)	1✔
115	coverage_gaps: List[str] = Field(	1✔
116	default_factory=list, description="Identified gaps in coverage"
117	)	1✔
118	coverage_patch: Optional[str] = Field(
NEW 119	None, description="Patch to fix coverage gaps"	×
120	)
121	patches: List[Patch] = Field(default_factory=list, description="Patches applied")
122	after_qa: str = Field(description="Summary after QA process")
123	diff_summary: Optional[str] = Field(None, description="Summary of changes made")
124	qa_status: Literal["PASS", "FAIL"] = Field(description="QA status")
125
126
127	class ModuleFinal(BaseModel):
128	"""Final summary and bullets for a module."""
129
130	summary: str = Field(description="Summary of the module's outcomes/activities")
131	bullets: List[str] = Field(
NEW 132	default_factory=list,	×
NEW 133	description="Key points or main sentiments from the module",	×
134	)
135		1✔
136
NEW 137	class PhaseModule(BaseModel):	×
NEW 138	"""Module within a phase section."""	×
NEW 139		×
140	module_name: str = Field(description="Name of the module")
141	module_id: int = Field(description="ID of the module")
142	status: Literal["past", "current", "upcoming"] = Field(description="Module status")	1✔
143	debug: Optional[ModuleDebug] = Field(None, description="Debug information")
144	final: ModuleFinal = Field(description="Summary and key points for the module")
145		1✔
146		1✔
147	class PhaseSection(BaseModel):
148	"""A phase section containing modules."""
149		1✔
150	modules: List[PhaseModule] = Field(
151	default_factory=list, description="Modules in this phase"
152	)	1✔
153
154
155	class Phases(BaseModel):
156	"""All phase sections."""
157
158	past: PhaseSection = Field(default_factory=PhaseSection)
159	current: PhaseSection = Field(default_factory=PhaseSection)
160	upcoming: PhaseSection = Field(default_factory=PhaseSection)
161
162
163	class GeneralInfo(BaseModel):
164	"""General information about the project."""
165
166	summary: str = Field(description="General summary of the entire project")
167	goals: List[str] = Field(
168	default_factory=list, description="Overall goals of the project"
169	)
170
171
172	class ProjectSummaryResponse(BaseModel):
173	"""Response model for complete project summarization."""
174
175	title: str = Field(default="Summary of participation")
176	general_info: GeneralInfo = Field(description="General project information")
177	phases: Phases = Field(description="All phase sections with their modules")
178
179
180	class DocumentInputItem(BaseModel):
181	"""Single document input item with handle and URL."""
182
183	handle: str = Field(description="Unique identifier/handle for the document")
184	url: str = Field(description="URL of the document")
185
186	def is_image(self) -> bool:
187	"""Check if the URL points to an image file."""
188	image_extensions = (
189	".jpg",
190	".jpeg",
191	".png",
192	".gif",
193	".webp",
194	".mpo",
195	".heif",
196	".avif",
197	".bmp",
198	".tiff",
199	".tif",
200	)
201	url_lower = self.url.lower()
202	return any(url_lower.endswith(ext) for ext in image_extensions)
203
204	def is_document(self) -> bool:
205	"""Check if the URL points to a document file (non-image)."""
206	document_extensions = (".pdf", ".doc", ".docx", ".txt", ".rtf", ".odt")
207	url_lower = self.url.lower()
208	return any(url_lower.endswith(ext) for ext in document_extensions)
209
210
211	class DocumentSummaryItem(BaseModel):
212	"""Response model for a single document summary with handle."""
213
214	handle: str = Field(description="Unique identifier/handle for the document")
215	summary: str = Field(description="Summary of the document content")
216
217
218	class DocumentSummaryResponse(BaseModel):
219	"""Response model for multiple document summaries."""
220
221	documents: list[DocumentSummaryItem] = Field(
222	default_factory=list,
223	description="List of document summaries, each with handle and summary",
224	)

liqd / roots / 22072536647

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