add more skill abbreviation

utilities/skills.py

@@ -28,9 +28,10 @@ FRONTEND = {
 
 # --- APIs & Architecture ---
 API_ARCH = {
-    "rest api", "restful api", "graphql", "grpc", "websocket",
+    "rest api", "restful api", "rest apis", "restful apis",
+    "graphql", "grpc", "websocket",
     "microservices", "microservice", "event driven", "message queue",
-    "api gateway", "api",
+    "api gateway", "api", "apis",
 }
 
 # --- Databases ---
@@ -60,6 +61,8 @@ ML_AI = {
     "data analysis", "data analytics", "data science",
     "statistical analysis", "statistics", "probability",
     "a/b testing", "hypothesis testing",
+    # Common abbreviations / shorthands
+    "ml", "ai", "dl", "cv",
 }
 
 # --- Data Engineering ---
@@ -153,9 +156,12 @@ def extract_skills(text: str) -> set:
                 break
             end = idx + len(skill)
 
-            # Boundary check — skill must be a whole word / phrase
+            # Boundary check — skill must start and end on a word boundary.
+            # Allow a trailing 's' so "rest apis" matches "rest api" etc.
             before_ok = (idx == 0 or cleaned[idx - 1] == ' ')
-            after_ok  = (end == len(cleaned) or cleaned[end] == ' ')
+            after_char = cleaned[end] if end < len(cleaned) else ' '
+            after_ok  = (after_char == ' ' or after_char == 's' and
+                         (end + 1 == len(cleaned) or cleaned[end + 1] == ' '))
 
             if before_ok and after_ok:
                 span = set(range(idx, end))
@@ -169,6 +175,47 @@ def extract_skills(text: str) -> set:
     return found
 
 
+# ---------------------------------------------------------------------------
+# Alias map — resume variant → canonical JD term
+# If a resume has any alias, it counts as having the canonical skill.
+# ---------------------------------------------------------------------------
+SKILL_ALIASES: dict[str, str] = {
+    # ML / AI shorthands
+    "ml":           "machine learning",
+    "ai":           "machine learning",
+    "dl":           "deep learning",
+    "cv":           "computer vision",
+    # API plurals / variants — chain: fastapi/rest apis → rest api → api
+    "apis":         "api",
+    "rest apis":    "rest api",
+    "restful apis": "restful api",
+    "rest api":     "api",
+    "restful api":  "api",
+    "api gateway":  "api",
+    "fastapi":      "api",
+    "grpc":         "api",
+    "graphql":      "api",
+}
+
+
+def _expand_with_aliases(skills: set) -> set:
+    """
+    Transitively expand aliases until no new terms are added.
+    e.g. fastapi → api; rest apis → rest api → api
+    """
+    expanded = set(skills)
+    while True:
+        new_terms = {
+            SKILL_ALIASES[s]
+            for s in expanded
+            if s in SKILL_ALIASES and SKILL_ALIASES[s] not in expanded
+        }
+        if not new_terms:
+            break
+        expanded |= new_terms
+    return expanded
+
+
 def extract_required_skills_from_jd(jd_text: str) -> dict:
     """Return JD skills with mention-frequency as an importance signal."""
     skills = extract_skills(jd_text)
@@ -177,18 +224,18 @@ def extract_required_skills_from_jd(jd_text: str) -> dict:
 
 
 def extract_resume_skills(resume_text: str) -> set:
-    return extract_skills(resume_text)
+    return _expand_with_aliases(extract_skills(resume_text))
 
 
 def find_missing_skills(resume_text: str, jd_text: str) -> list:
-    jd_skills   = set(extract_required_skills_from_jd(jd_text).keys())
-    resume_skills = extract_resume_skills(resume_text)
+    jd_skills     = set(extract_required_skills_from_jd(jd_text).keys())
+    resume_skills = extract_resume_skills(resume_text)   # already alias-expanded
     return [s for s in jd_skills if s not in resume_skills]
 
 
 def calculate_skill_overlap(resume_text: str, jd_text: str) -> float:
     jd_skills     = set(extract_required_skills_from_jd(jd_text).keys())
-    resume_skills = extract_resume_skills(resume_text)
+    resume_skills = extract_resume_skills(resume_text)   # already alias-expanded
     if not jd_skills:
         return 0.0
     return round(len(jd_skills & resume_skills) / len(jd_skills) * 100, 2)