Merge pull request #22 from Riminder/feat/add-scoring-searching-profile-validation

Feat: add validation for profile searching & scoring in utils

Author: corentin-hrflow

hrflow/schemas.py

@@ -241,6 +241,12 @@ class Experience(BaseModel):
     tasks: t.Optional[t.List[GeneralEntitySchema]] = Field(
         None, description="List of tasks of the Experience."
     )
+    languages: t.Optional[t.List[GeneralEntitySchema]] = Field(
+        None, description="List of spoken languages of the profile"
+    )
+    interests: t.Optional[t.List[GeneralEntitySchema]] = Field(
+        None, description="List of interests of the Experience."
+    )
 
 
 class Education(BaseModel):
@@ -274,6 +280,12 @@ class Education(BaseModel):
     tasks: t.Optional[t.List[GeneralEntitySchema]] = Field(
         None, description="List of tasks of the Education."
     )
+    languages: t.Optional[t.List[GeneralEntitySchema]] = Field(
+        None, description="List of spoken languages of the profile"
+    )
+    interests: t.Optional[t.List[GeneralEntitySchema]] = Field(
+        None, description="List of interests of the Experience."
+    )
 
 
 class Attachment(BaseModel):

hrflow/utils/__init__.py

@@ -1,3 +1,5 @@
 # noqa: F401
 from .evaluation import generate_parsing_evaluation_report
+from .scoring import is_valid_for_scoring
+from .searching import is_valid_for_searching
 from .storing import get_all_jobs, get_all_profiles

hrflow/utils/scoring.py

@@ -0,0 +1,75 @@
+import typing as t
+
+from ..schemas import Education, Experience, HrFlowProfile
+from .searching import is_valid_for_searching
+
+
+def is_valid_experiences_for_scoring(
+    experience_list: t.Optional[t.List[Experience]],
+) -> bool:
+    """
+    Check if a list of experiences is valid for scoring
+
+    Args:
+        experience_list:        <list>
+                                List of experiences to check
+    Return:
+        <bool>                  True if the list of experiences is valid for
+                                scoring, False otherwise
+    """
+    for experience in experience_list:
+        if experience.title:
+            return True
+
+    return False
+
+
+def is_valid_educations_for_scoring(
+    education_list: t.Optional[t.List[Education]],
+) -> bool:
+    """
+    Check if a list of educations is valid for scoring
+
+    Args:
+        education_list:         <list>
+                                List of educations to check
+    Return:
+        <bool>                  True if the list of educations is valid for
+                                scoring, False otherwise
+    """
+    for education in education_list:
+        if education.title or education.school:
+            return True
+    return False
+
+
+def is_valid_for_scoring(
+    profile: t.Union[t.Dict, HrFlowProfile],
+) -> bool:
+    """
+    Check if a profile is valid for scoring
+
+    Based on the following schemas https://developers.hrflow.ai/docs/profiles-scoring
+
+    Args:
+        client:                 <Hrflow>
+                                Hrflow client
+        profile:                <dict> or <HrFlowProfile>
+                                Profile to check
+    Return:
+        <bool>                  True if the profile is valid for scoring,
+                                False otherwise
+    """
+    if isinstance(profile, dict):
+        profile = HrFlowProfile.parse_obj(profile)
+
+    if not isinstance(profile, HrFlowProfile):
+        raise ValueError("profile must be a dict or a HrFlowProfile object")
+
+    return is_valid_for_searching(profile) and (
+        is_valid_experiences_for_scoring(profile.experiences)
+        or is_valid_educations_for_scoring(profile.educations)
+        or bool(profile.summary)
+        or bool(profile.skills)
+        or bool(profile.tasks)
+    )

hrflow/utils/searching.py

@@ -0,0 +1,67 @@
+import typing as t
+
+from ..schemas import HrFlowProfile, ProfileInfo
+
+
+def is_valid_info_for_searching(info: ProfileInfo) -> bool:
+    """
+    Check if the info part of a profile is valid for searching
+
+    Based on the following schemas https://developers.hrflow.ai/docs/profiles-searching
+
+    Args:
+        info:                   <ProfileInfo>
+                                Info part of the profile
+    """
+    if not isinstance(info, ProfileInfo):
+        raise ValueError("info must be a ProfileInfo object")
+
+    first_name_score = 1 if info.first_name else 0
+    last_name_score = 1 if info.last_name else 0
+    phone_score = 1 if info.phone else 0
+    date_birth_score = 1 if info.date_birth else 0
+    gender_score = 1 if info.gender else 0
+    summary_score = 1 if info.summary else 0
+    urls_score = 1 if info.urls else 0
+    location_score = 1 if info.location else 0
+
+    info_score = (
+        first_name_score
+        + last_name_score
+        + phone_score
+        + date_birth_score
+        + gender_score
+        + summary_score
+        + urls_score
+        + location_score
+    )
+    info_score = info_score / 8
+    has_person = info.first_name and info.last_name
+
+    return info.email or has_person or info_score >= 0.5
+
+
+def is_valid_for_searching(
+    profile: t.Union[t.Dict, HrFlowProfile],
+) -> bool:
+    """
+    Check if a profile is valid for searching
+
+    Based on the following schemas https://developers.hrflow.ai/docs/profiles-searching
+
+    Args:
+        client:                 <Hrflow>
+                                Hrflow client
+        profile:                <dict> or <HrFlowProfile>
+                                Profile to check
+    Return:
+        <bool>                  True if the profile is valid for searching,
+                                False otherwise
+    """
+    if isinstance(profile, dict):
+        profile = HrFlowProfile.parse_obj(profile)
+
+    if not isinstance(profile, HrFlowProfile):
+        raise ValueError("profile must be a dict or a HrFlowProfile object")
+
+    return is_valid_info_for_searching(profile.info) and bool(profile.text)