Merge pull request #51 from olp-cs/upgrade-to-pydantic-v2

Upgrade to Pydantic 2.8.2

Author: woodthom2

pyproject.toml

@@ -36,7 +36,7 @@ classifiers=[
 # core dependencies of harmony
 # this set should be kept minimal!
 dependencies = [
-    "pydantic==1.10.7; python_version <= '3.12'",
+    "pydantic==2.8.2; python_version <= '3.12'",
     "pandas==2.0.0; python_version <= '3.12'",
     "tika==2.6.0; python_version <= '3.12'",
     "lxml==4.9.2; python_version <= '3.12'",

requirements.txt

@@ -1,4 +1,4 @@
-pydantic==1.10.7
+pydantic==2.8.2
 pandas==2.0.0
 tika==2.6.0
 lxml==4.9.2

src/harmony/examples.py

@@ -208,7 +208,7 @@ def match_instruments_with_function(
                     ctrs[all_questions[idx].instrument_id] = Counter()
                 for topic in mhc_all_metadatas[mhc_item_idx]["topics"]:
                     ctrs[all_questions[idx].instrument_id][topic] += 1
-                all_questions[idx].nearest_match_from_mhc_auto = mhc_questions[mhc_item_idx].dict()
+                all_questions[idx].nearest_match_from_mhc_auto = mhc_questions[mhc_item_idx].model_dump()
                 strength_of_match = similarities_mhc[idx, mhc_item_idx]
                 all_questions[idx].topics_strengths = {topic: float(strength_of_match)}
 

src/harmony/matching/matcher.py

@@ -29,7 +29,7 @@
 
 from harmony.schemas.requests.text import Instrument
 from harmony.schemas.requests.text import Question
-from pydantic import BaseModel, Field
+from pydantic import BaseModel, Field, RootModel
 
 
 class MatchResponse(BaseModel):
@@ -42,8 +42,8 @@ class MatchResponse(BaseModel):
     )
 
 
-class InstrumentList(BaseModel):
-    __root__: List[Instrument]
+class InstrumentList(RootModel):
+    root: List[Instrument]
 
 
 class CacheResponse(BaseModel):

src/harmony/schemas/requests/text.py

@@ -52,7 +52,7 @@ def import_instrument_into_harmony_web(instrument: Instrument, harmony_fe_base_u
     @param harmony_fe_base_url: The base URL of the React app front end, defaulting to the web Harmony front end at harmonydata.ac.uk
     @return: a URL which you can click which will take you to the browser.
     """
-    instrument_serialised_as_json = json.dumps(instrument.dict())
+    instrument_serialised_as_json = json.dumps(instrument.model_dump())
     instrument_json_b64_encoded_bytes = base64.urlsafe_b64encode(instrument_serialised_as_json.encode('utf-8'))
     instrument_json_b64_encoded_str = instrument_json_b64_encoded_bytes.decode("utf-8")
 

src/harmony/schemas/responses/text.py

@@ -33,7 +33,7 @@
 from harmony import convert_excel_to_instruments
 from harmony.schemas.requests.text import RawFile
 
-xlsx_gad_7_2_questions = RawFile.parse_obj({
+xlsx_gad_7_2_questions = RawFile.model_validate({
     "file_id": "1d66bce4b80c4b0eaefe33f00cddedef",
     "file_name": "GAD-7.xlsx",
     "file_type": "xlsx",

src/harmony/util/instrument_helper.py

@@ -33,7 +33,7 @@
 
 sys.path.append("../src")
 
-xlsx_gad_7_2_questions = RawFile.parse_obj({
+xlsx_gad_7_2_questions = RawFile.model_validate({
     "file_id": "1d66bce4b80c4b0eaefe33f00cddedef",
     "file_name": "GAD-7.xlsx",
     "file_type": "xlsx",

tests/test_convert_excel_openpyxl.py

@@ -34,7 +34,7 @@
 from harmony.schemas.requests.text import RawFile
 from harmony import download_models
 
-pdf_gad_7_2_questions = RawFile.parse_obj({
+pdf_gad_7_2_questions = RawFile.model_validate({
     "file_id": "d39f31718513413fbfc620c6b6135d0c",
     "file_name": "GAD-7.pdf",
     "file_type": "pdf",

tests/test_convert_excel_xlsxwriter.py

@@ -33,7 +33,7 @@
 from harmony import convert_text_to_instruments
 from harmony.schemas.requests.text import RawFile
 
-txt_gad_7_2_questions = RawFile.parse_obj({
+txt_gad_7_2_questions = RawFile.model_validate({
     "file_id": "d39f31718513413fbfc620c6b6135d0c",
     "file_name": "GAD-7.txt",
     "file_type": "txt",

tests/test_convert_pdf.py

@@ -43,7 +43,7 @@
         question_text="Durante as últimas 2 semanas, com que freqüência você foi incomodado/a pelos problemas abaixo?")]
 instrument_pt = Instrument(questions=questions_pt, language="pt")
 
-instrument_1 = Instrument.parse_obj({
+instrument_1 = Instrument.model_validate({
     "file_id": "fd60a9a64b1b4078a68f4bc06f20253c",
     "instrument_id": "7829ba96f48e4848abd97884911b6795",
     "instrument_name": "GAD-7 English",
@@ -80,7 +80,7 @@
 }
 )
 
-instrument_2 = Instrument.parse_obj({
+instrument_2 = Instrument.model_validate({
     "file_id": "fd60a9a64b1b4078a68f4bc06f20253c",
     "instrument_id": "7829ba96f48e4848abd97884911b6795",
     "instrument_name": "GAD-7 Portuguese",

tests/test_convert_text.py

@@ -35,7 +35,7 @@
 from harmony import download_models
 
 
-pdf_empty_table = RawFile.parse_obj({
+pdf_empty_table = RawFile.model_validate({
     "file_id": "d39f31718513413fbfc620c6b6135d0c",
     "file_name": "GAD-7.pdf",
     "file_type": "pdf",
@@ -44,7 +44,7 @@
     "content":""
 })
 
-pdf_non_empty_table = RawFile.parse_obj({
+pdf_non_empty_table = RawFile.model_validate({
     "file_id": "d39f31718513413fbfc620c6b6135d0c",
     "file_name": "GAD-7.pdf",
     "file_type": "pdf",