import polars as pl from pathlib import Path import pandas as pd from typing import Union import json import re def extract_voice_label(html_str: str) -> str: """ Extract voice label from HTML string and convert to short format. Parameters: html_str (str): HTML string containing voice label in format "Voice N" Returns: str: Voice label in format "VN" (e.g., "V14") Example: >>> extract_voice_label('Voice 14
...') 'V14' """ match = re.search(r'Voice (\d+)', html_str) return f"V{match.group(1)}" if match else None def extract_qid(val): """Extracts the 'ImportId' from a string representation of a dictionary.""" if isinstance(val, str) and val.startswith('{') and val.endswith('}'): val = eval(val) return val['ImportId'] class JPMCSurvey: """Class to handle JPMorgan Chase survey data.""" def __init__(self, data_path: Union[str, Path], qsf_path: Union[str, Path]): if isinstance(data_path, str): data_path = Path(data_path) if isinstance(qsf_path, str): qsf_path = Path(qsf_path) self.data_filepath = data_path self.qsf_filepath = qsf_path self.qid_descr_map = self._extract_qid_descr_map() self.qsf:dict = self._load_qsf() def _extract_qid_descr_map(self) -> dict: """Extract mapping of Qualtrics ImportID to Question Description from results file.""" if '1_1-16-2026' in self.data_filepath.as_posix(): df_questions = pd.read_csv(self.data_filepath, nrows=1) df_questions return df_questions.iloc[0].to_dict() else: # First row contains Qualtrics Editor question names (ie 'B_VOICE SEL. 18-8') # Second row which contains the question content # Third row contains the Export Metadata (ie '{"ImportId":"startDate","timeZone":"America/Denver"}') df_questions = pd.read_csv(self.data_filepath, nrows=2) # transpose df_questions df_questions = df_questions.T.reset_index() df_questions.columns = ['QName', 'Description', 'export_metadata'] df_questions['ImportID'] = df_questions['export_metadata'].apply(extract_qid) df_questions = df_questions[['ImportID', 'QName', 'Description']] # return dict as {ImportID: [QName, Description]} return df_questions.set_index('ImportID')[['QName', 'Description']].T.to_dict() def _load_qsf(self) -> dict: """Load QSF file to extract question metadata if needed.""" with open(self.qsf_filepath, 'r', encoding='utf-8') as f: qsf_data = json.load(f) return qsf_data def _get_qsf_question_by_QID(self, QID: str) -> dict: """Get question metadata from QSF using the Question ID.""" q_elem = [elem for elem in self.qsf['SurveyElements'] if elem['PrimaryAttribute'] == QID] if len(q_elem) == 0: raise ValueError(f"SurveyElement with 'PrimaryAttribute': '{QID}' not found in QSF.") if len(q_elem) > 1: raise ValueError(f"Multiple SurveyElements with 'PrimaryAttribute': '{QID}' found in QSF: \n{q_elem}") return q_elem[0] def load_data(self) -> pl.LazyFrame: """ Load CSV where column headers are in row 3 as dict strings with ImportId. The 3rd row contains metadata like '{"ImportId":"startDate","timeZone":"America/Denver"}'. This function extracts the ImportId from each column and uses it as the column name. Parameters: file_path (Path): Path to the CSV file to load. Returns: pl.LazyFrame: Polars LazyFrame with ImportId as column names. """ if '1_1-16-2026' in self.data_filepath.as_posix(): raise NotImplementedError("This method does not support the '1_1-16-2026' export format.") # Read the 3rd row (index 2) which contains the metadata dictionaries # Use header=None to get raw values instead of treating them as column names df_meta = pd.read_csv(self.data_filepath, nrows=1, skiprows=2, header=None) # Extract ImportIds from each column value in this row new_columns = [extract_qid(val) for val in df_meta.iloc[0]] # Now read the actual data starting from row 4 (skip first 3 rows) df = pl.read_csv(self.data_filepath, skip_rows=3) # Rename columns with the extracted ImportIds df.columns = new_columns return df.lazy() def _get_subset(self, q: pl.LazyFrame, QIDs, rename_cols=True) -> pl.LazyFrame: """Extract subset of data based on specific questions.""" if not rename_cols: return q.select(QIDs) rename_dict = {qid: self.qid_descr_map[qid]['QName'] for qid in QIDs if qid in self.qid_descr_map} return q.select(QIDs).rename(rename_dict) def get_demographics(self, q: pl.LazyFrame) -> pl.LazyFrame: """Extract columns containing the demographics. Renames columns using qid_descr_map if provided. """ QIDs = ['QID1', 'QID2', 'QID3', 'QID4', 'QID13', 'QID14', 'QID15', 'QID16', 'QID17', 'Consumer'] return self._get_subset(q, QIDs), None def get_top_8_traits(self, q: pl.LazyFrame) -> pl.LazyFrame: """Extract columns containing the top 8 characteristics are most important for this Chase virtual assistant to have. Returns subquery that can be chained with other polars queries. """ QIDs = ['QID25'] return self._get_subset(q, QIDs, rename_cols=False).rename({'QID25': 'Top_8_Traits'}), None def get_top_3_traits(self, q: pl.LazyFrame) -> pl.LazyFrame: """Extract columns containing the top 3 characteristics that the Chase virtual assistant should prioritize. Returns subquery that can be chained with other polars queries. """ QIDs = ['QID26_0_GROUP'] return self._get_subset(q, QIDs, rename_cols=False).rename({'QID26_0_GROUP': 'Top_3_Traits'}), None def get_character_ranking(self, q: pl.LazyFrame) -> pl.LazyFrame: """Extract columns containing the ranking of characteristics for the Chase virtual assistant. Returns subquery that can be chained with other polars queries. """ # Requires QSF to map "Character Ranking_2" to the actual character cfg = self._get_qsf_question_by_QID('QID27')['Payload'] QIDs_map = {f'QID27_{v}': cfg['VariableNaming'][k] for k,v in cfg['RecodeValues'].items()} QIDs_rename = {qid: f'Character_Ranking_{QIDs_map[qid].replace(" ", "_")}' for qid in QIDs_map} return self._get_subset(q, list(QIDs_rename.keys()), rename_cols=False).rename(QIDs_rename), None def get_18_8_3(self, q: pl.LazyFrame) -> pl.LazyFrame: """Extract columns containing the 18-8-3 feedback for the Chase virtual assistant. Returns subquery that can be chained with other polars queries. """ QIDs = ['QID29', 'QID101', 'QID36_0_GROUP'] rename_dict = { 'QID29': '18-8_Set-A', 'QID101': '18-8_Set-B', 'QID36_0_GROUP': '8-3_Ranked' } return self._get_subset(q, QIDs, rename_cols=False).rename(rename_dict), None def get_voice_scale_1_10(self, q: pl.LazyFrame) -> pl.LazyFrame: """Extract columns containing the Voice Scale 1-10 ratings for the Chase virtual assistant. Returns subquery that can be chained with other polars queries. """ QIDs_map = {} for qid, val in self.qid_descr_map.items(): if 'Scale 1-10_1' in val['QName']: # Convert "Voice 16 Scale 1-10_1" to "Scale_1_10__Voice_16" QIDs_map[qid] = f"Voice_Scale_1_10__V{val['QName'].split()[1]}" return self._get_subset(q, list(QIDs_map.keys()), rename_cols=False).rename(QIDs_map), None def get_ss_green_blue(self, q: pl.LazyFrame) -> pl.LazyFrame: """Extract columns containing the SS Green/Blue ratings for the Chase virtual assistant. Returns subquery that can be chained with other polars queries. """ cfg = self._get_qsf_question_by_QID('QID35')['Payload'] QIDs_map = {} choices_map = {} for qid, val in self.qid_descr_map.items(): if 'SS Green-Blue' in val['QName']: cfg = self._get_qsf_question_by_QID(qid.split('_')[0])['Payload'] # ie: "V14 SS Green-Blue_1" qname_parts = val['QName'].split() voice = qname_parts[0] trait_num = qname_parts[-1].split('_')[-1] QIDs_map[qid] = f"SS_Green_Blue__{voice}__Choice_{trait_num}" choices_map[f"SS_Green_Blue__{voice}__Choice_{trait_num}"] = cfg['Choices'][trait_num]['Display'] return self._get_subset(q, list(QIDs_map.keys()), rename_cols=False).rename(QIDs_map), choices_map def get_top_3_voices(self, q: pl.LazyFrame) -> pl.LazyFrame: """Extract columns containing the top 3 voice choices for the Chase virtual assistant. Returns subquery that can be chained with other polars queries. """ QIDs_map = {} cfg36 = self._get_qsf_question_by_QID('QID36')['Payload'] choice_voice_map = {k: extract_voice_label(v['Display']) for k,v in cfg36['Choices'].items()} for qid, val in self.qid_descr_map.items(): if 'Rank Top 3 Voices' in val['QName']: cfg = self._get_qsf_question_by_QID(qid.split('_')[0])['Payload'] voice_num = val['QName'].split('_')[-1] # Validate that the DynamicChoices Locator is as expected if cfg['DynamicChoices']['Locator'] != r"q://QID36/ChoiceGroup/SelectedChoicesInGroup/1": raise ValueError(f"Unexpected DynamicChoices Locator for QID '{qid}': {cfg['DynamicChoices']['Locator']}") # extract the voice from the QID36 config voice = choice_voice_map[voice_num] # Convert "Top 3 Voices_1" to "Top_3_Voices__V14" QIDs_map[qid] = f"Top_3_Voices_ranking__{voice}" return self._get_subset(q, list(QIDs_map.keys()), rename_cols=False).rename(QIDs_map), None def get_ss_orange_red(self, q: pl.LazyFrame) -> pl.LazyFrame: """Extract columns containing the SS Orange/Red ratings for the Chase virtual assistant. Returns subquery that can be chained with other polars queries. """ cfg = self._get_qsf_question_by_QID('QID40')['Payload'] QIDs_map = {} choices_map = {} for qid, val in self.qid_descr_map.items(): if 'SS Orange-Red' in val['QName']: cfg = self._get_qsf_question_by_QID(qid.split('_')[0])['Payload'] # ie: "V14 SS Orange-Red_1" qname_parts = val['QName'].split() voice = qname_parts[0] trait_num = qname_parts[-1].split('_')[-1] QIDs_map[qid] = f"SS_Orange_Red__{voice}__Choice_{trait_num}" choices_map[f"SS_Orange_Red__{voice}__Choice_{trait_num}"] = cfg['Choices'][trait_num]['Display'] return self._get_subset(q, list(QIDs_map.keys()), rename_cols=False).rename(QIDs_map), choices_map def get_character_refine(self, q: pl.LazyFrame) -> pl.LazyFrame: """Extract columns containing the character refine feedback for the Chase virtual assistant. Returns subquery that can be chained with other polars queries. """ QIDs = ['QID29', 'QID101', 'QID36_0_GROUP'] rename_dict = { 'QID29': '18-8_Set-A', 'QID101': '18-8_Set-B', 'QID36_0_GROUP': '8-3_Ranked' }