Spaces:
Running
Running
File size: 14,363 Bytes
dffabb7 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 | #!/usr/bin/env python3
"""
src/summarizer.py — Match Explanation Generator
Generates ≤80-word explanations in EN or FR explaining why a tender matches a profile.
Uses template-based generation (CPU-only, no LLM dependency required).
"""
import random
# ─── English Templates ────────────────────────────────────────────────────────
EN_TEMPLATES = [
(
"{org_name} matches **{tender_title}** (score: {score:.2f}). "
"This {sector} grant from {tender_region} aligns with your operations in {country}. "
"The available funding of USD {budget:,} fits your budget range. "
"Deadline: {deadline}. "
"Sector overlap and {tfidf_pct}% content similarity drive this ranking."
),
(
"**{tender_title}** is ranked #{rank} for {org_name}. "
"Sector: {sector} ✓. Budget: USD {budget:,}. Deadline: {deadline}. "
"Your needs in {needs_snippet} closely match this tender's objectives. "
"Score breakdown — similarity: {tfidf_pct}%, sector: {sector_pct}%, budget: {budget_pct}%."
),
(
"This {sector} opportunity suits {org_name} because your profile in {country} aligns "
"with the tender's focus on {region_phrase}. "
"Budget of USD {budget:,} is within reach. Apply before {deadline}. "
"Composite match score: {score:.2f}/1.00."
),
]
# ─── French Templates ─────────────────────────────────────────────────────────
FR_TEMPLATES = [
(
"{org_name} correspond à **{tender_title}** (score : {score:.2f}). "
"Cette subvention {sector} en {tender_region} s'aligne avec vos activités en {country}. "
"Le financement disponible de USD {budget:,} correspond à votre capacité budgétaire. "
"Date limite : {deadline}. "
"La correspondance sectorielle et {tfidf_pct}% de similarité de contenu motivent ce classement."
),
(
"**{tender_title}** est classé #{rank} pour {org_name}. "
"Secteur : {sector} ✓. Budget : USD {budget:,}. Date limite : {deadline}. "
"Vos besoins en {needs_snippet} correspondent étroitement aux objectifs de cet appel. "
"Détail du score — similarité : {tfidf_pct}%, secteur : {sector_pct}%, budget : {budget_pct}%."
),
(
"Cette opportunité {sector} convient à {org_name} car votre profil en {country} s'aligne "
"avec l'appel ciblant {region_phrase}. "
"Le budget de USD {budget:,} est accessible. Déposez votre candidature avant le {deadline}. "
"Score composite : {score:.2f}/1.00."
),
]
SECTOR_PHRASES_EN = {
"agritech": "digital agriculture and farming innovation",
"healthtech": "health technology and community health services",
"cleantech": "clean and renewable energy solutions",
"edtech": "digital education and offline learning",
"fintech": "digital finance and financial inclusion",
"wastetech": "waste management and circular economy",
"general": "general development and innovation",
}
SECTOR_PHRASES_FR = {
"agritech": "l'agriculture numérique et l'innovation agricole",
"healthtech": "la technologie de santé et les services de santé communautaire",
"cleantech": "les solutions d'énergie propre et renouvelable",
"edtech": "l'éducation numérique et l'apprentissage hors-ligne",
"fintech": "la finance numérique et l'inclusion financière",
"wastetech": "la gestion des déchets et l'économie circulaire",
"general": "le développement général et l'innovation",
}
REGION_PHRASES_EN = {
"East Africa": "East African markets",
"West Africa": "West African communities",
"Central Africa": "Central African regions",
"Southern Africa": "Southern African areas",
"Africa": "pan-African initiatives",
}
REGION_PHRASES_FR = {
"East Africa": "les marchés d'Afrique de l'Est",
"West Africa": "les communautés d'Afrique de l'Ouest",
"Central Africa": "les régions d'Afrique Centrale",
"Southern Africa": "les zones d'Afrique Australe",
"Africa": "les initiatives panafricaines",
}
def _truncate_to_words(text: str, max_words: int = 80) -> str:
"""Truncate text to max_words, ending at a sentence boundary if possible."""
words = text.split()
if len(words) <= max_words:
return text
truncated = " ".join(words[:max_words])
# Try to end at last sentence
for punct in [".", "!", "?"]:
idx = truncated.rfind(punct)
if idx > len(truncated) // 2:
return truncated[:idx + 1]
return truncated + "..."
def generate_summary(
profile: dict,
tender: dict,
rank: int,
score: float,
breakdown: dict,
language: str = "en",
max_words: int = 80,
) -> str:
"""
Generate a ≤80-word explanation of why this tender matches the profile.
Args:
profile: business profile dict
tender: matched tender dict
rank: rank position (1–5)
score: composite match score (0–1)
breakdown: dict with tfidf_similarity, sector_match, budget_score, urgency_score
language: "en" or "fr"
max_words: word limit (default 80)
Returns:
Formatted explanation string
"""
lang = language if language in ["en", "fr"] else "en"
# Derived values
tfidf_pct = int(breakdown.get("tfidf_similarity", 0) * 100)
sector_pct = int(breakdown.get("sector_match", 0) * 100)
budget_pct = int(breakdown.get("budget_score", 0) * 100)
urgency_pct = int(breakdown.get("urgency_score", 0) * 100)
sector = tender.get("sector", "general")
region = tender.get("region", "Africa")
needs_text = profile.get("needs_text", "")
needs_snippet = " ".join(needs_text.split()[:6]) + "..." if needs_text else "various areas"
if lang == "fr":
templates = FR_TEMPLATES
region_phrase = REGION_PHRASES_FR.get(region, "les régions africaines")
else:
templates = EN_TEMPLATES
region_phrase = REGION_PHRASES_EN.get(region, "African regions")
template = templates[rank % len(templates)]
summary = template.format(
org_name=profile.get("name", "Your organization"),
tender_title=tender.get("title", "This Tender"),
score=score,
sector=sector,
country=profile.get("country", "your country"),
budget=tender.get("budget", 0),
deadline=tender.get("deadline", "TBD"),
tfidf_pct=tfidf_pct,
sector_pct=sector_pct,
budget_pct=budget_pct,
urgency_pct=urgency_pct,
rank=rank,
needs_snippet=needs_snippet,
tender_region=region,
region_phrase=region_phrase,
)
return _truncate_to_words(summary, max_words)
def generate_summary_md(
profile: dict,
matches: list,
language: str = "en",
) -> str:
"""
Generate a complete markdown summary file for all matches of a profile.
Args:
profile: business profile dict
matches: list of ranked tender dicts (from ranker.rank())
language: "en" or "fr"
Returns:
Full markdown string
"""
lang = language if language in ["en", "fr"] else "en"
lines = []
if lang == "fr":
lines.append(f"# Correspondances de Subventions — {profile.get('name', 'Profil')}")
lines.append(f"\n**Profil :** {profile.get('name')} | **Secteur :** {profile.get('sector')} | **Pays :** {profile.get('country')}")
lines.append(f"\n**Besoins :** {profile.get('needs_text', '')}\n")
lines.append("---\n")
lines.append("## Top 5 Appels à Candidatures\n")
else:
lines.append(f"# Grant Matches — {profile.get('name', 'Profile')}")
lines.append(f"\n**Profile:** {profile.get('name')} | **Sector:** {profile.get('sector')} | **Country:** {profile.get('country')}")
lines.append(f"\n**Needs:** {profile.get('needs_text', '')}\n")
lines.append("---\n")
lines.append("## Top 5 Matched Tenders\n")
for rank, match in enumerate(matches, 1):
score = match["score"]
breakdown = match["breakdown"]
summary = generate_summary(
profile=profile,
tender=match,
rank=rank,
score=score,
breakdown=breakdown,
language=lang,
)
if lang == "fr":
lines.append(f"### #{rank} — {match['title']}")
lines.append(f"**ID :** {match['tender_id']} | **Score :** {score:.4f} | **Langue :** {match['language'].upper()}")
lines.append(f"\n**Explication :**\n{summary}\n")
lines.append(f"**Détail du score :**")
lines.append(f"- Similarité TF-IDF : {breakdown['tfidf_similarity']:.3f}")
lines.append(f"- Correspondance sectorielle : {breakdown['sector_match']:.3f}")
lines.append(f"- Compatibilité budgétaire : {breakdown['budget_score']:.3f}")
lines.append(f"- Urgence deadline : {breakdown['urgency_score']:.3f}\n")
else:
lines.append(f"### #{rank} — {match['title']}")
lines.append(f"**ID:** {match['tender_id']} | **Score:** {score:.4f} | **Language:** {match['language'].upper()}")
lines.append(f"\n**Explanation:**\n{summary}\n")
lines.append(f"**Score Breakdown:**")
lines.append(f"- TF-IDF Similarity: {breakdown['tfidf_similarity']:.3f}")
lines.append(f"- Sector Match: {breakdown['sector_match']:.3f}")
lines.append(f"- Budget Compatibility: {breakdown['budget_score']:.3f}")
lines.append(f"- Deadline Urgency: {breakdown['urgency_score']:.3f}\n")
lines.append("---\n")
return "\n".join(lines)
def generate_individual_summary_md(
profile: dict,
match: dict,
rank: int,
language: str = "en",
disqualifier: str = "",
) -> str:
"""
Generate a single .md file for one (profile, tender) match pair.
Spec requires one .md per (profile, tender) match in summaries/.
Args:
profile: business profile dict
match: single ranked tender dict (from ranker.rank())
rank: rank position (1-based)
language: "en" or "fr"
disqualifier: pre-computed top disqualifier string
Returns:
Markdown string for this individual match
"""
from src.utils import format_budget
lang = language if language in ["en", "fr"] else "en"
score = match["score"]
breakdown = match["breakdown"]
tid = match["tender_id"]
summary_text = generate_summary(
profile=profile,
tender=match,
rank=rank,
score=score,
breakdown=breakdown,
language=lang,
)
budget_str = format_budget(match.get("budget", 0))
disq = disqualifier or "No major disqualifier identified."
if lang == "fr":
return (
f"# {match['title']}\n"
f"**Profil :** {profile.get('name')} | **ID :** {profile.get('id')} "
f"| **Langue :** {lang.upper()}\n\n"
"---\n\n"
f"## Résumé de Correspondance (#{rank})\n\n"
f"{summary_text}\n\n"
"---\n\n"
"## Détails\n\n"
"| Champ | Valeur |\n|-------|--------|\n"
f"| ID Appel | {tid} |\n"
f"| Score Composite | {score:.4f} |\n"
f"| Secteur | {match['sector']} |\n"
f"| Budget | {budget_str} |\n"
f"| Date Limite | {match['deadline']} |\n"
f"| Région | {match['region']} |\n"
f"| Langue du Document | {match['language'].upper()} |\n\n"
"## Détail du Score\n\n"
"| Composant | Score |\n|-----------|-------|\n"
f"| Similarité TF-IDF | {breakdown['tfidf_similarity']:.3f} |\n"
f"| Correspondance Sectorielle | {breakdown['sector_match']:.3f} |\n"
f"| Compatibilité Budgétaire | {breakdown['budget_score']:.3f} |\n"
f"| Urgence Deadline | {breakdown['urgency_score']:.3f} |\n\n"
f"## ⚠ Principal Facteur Disqualifiant\n\n{disq}\n"
)
else:
return (
f"# {match['title']}\n"
f"**Profile:** {profile.get('name')} | **ID:** {profile.get('id')} "
f"| **Language:** {lang.upper()}\n\n"
"---\n\n"
f"## Match Summary (#{rank})\n\n"
f"{summary_text}\n\n"
"---\n\n"
"## Details\n\n"
"| Field | Value |\n|-------|-------|\n"
f"| Tender ID | {tid} |\n"
f"| Composite Score | {score:.4f} |\n"
f"| Sector | {match['sector']} |\n"
f"| Budget | {budget_str} |\n"
f"| Deadline | {match['deadline']} |\n"
f"| Region | {match['region']} |\n"
f"| Document Language | {match['language'].upper()} |\n\n"
"## Score Breakdown\n\n"
"| Component | Score |\n|-----------|-------|\n"
f"| TF-IDF Similarity | {breakdown['tfidf_similarity']:.3f} |\n"
f"| Sector Match | {breakdown['sector_match']:.3f} |\n"
f"| Budget Compatibility | {breakdown['budget_score']:.3f} |\n"
f"| Deadline Urgency | {breakdown['urgency_score']:.3f} |\n\n"
f"## ⚠ Biggest Disqualifier\n\n{disq}\n"
)
if __name__ == "__main__":
# Quick test
profile = {
"id": "01", "name": "AgriGrow Rwanda", "sector": "agritech",
"country": "Rwanda", "budget_max": 50000,
"needs_text": "We need funding to scale our precision farming app.",
"languages": ["en"]
}
tender = {
"id": "T004", "title": "Digital Agriculture Innovation Grant",
"sector": "agritech", "budget": 50000, "deadline": "15 August 2025",
"region": "East Africa", "language": "en"
}
breakdown = {"tfidf_similarity": 0.45, "sector_match": 1.0, "budget_score": 1.0, "urgency_score": 0.65}
print("=== EN Summary ===")
print(generate_summary(profile, tender, 1, 0.78, breakdown, "en"))
print("\n=== FR Summary ===")
print(generate_summary(profile, tender, 1, 0.78, breakdown, "fr"))
|