Update process_interview.py

process_interview.py

@@ -328,12 +328,19 @@ def analyze_interviewee_voice(audio_path: str, utterances: List[Dict]) -> Dict:
         y, sr = librosa.load(audio_path, sr=16000)
         interviewee_utterances = [u for u in utterances if u['role'] == 'Interviewee']
         if not interviewee_utterances:
+            logger.warning("No interviewee utterances found")
             return {'error': 'No interviewee utterances found'}
         segments = []
         for u in interviewee_utterances:
             start = int(u['start'] * sr / 1000)
             end = int(u['end'] * sr / 1000)
-            segments.append(y[start:end])
+            if end > start and len(y[start:end]) > 0:
+                segments.append(y[start:end])
+            else:
+                logger.warning(f"Invalid segment for utterance: start={start}, end={end}")
+        if not segments:
+            logger.warning("No valid audio segments for voice analysis")
+            return {'error': 'No valid audio segments found'}
         total_duration = sum(u['prosodic_features']['duration'] for u in interviewee_utterances)
         total_words = sum(len(u['text'].split()) for u in interviewee_utterances)
         speaking_rate = total_words / total_duration if total_duration > 0 else 0
@@ -355,11 +362,11 @@ def analyze_interviewee_voice(audio_path: str, utterances: List[Dict]) -> Dict:
         jitter = np.mean(np.abs(np.diff(pitches))) / pitch_mean if len(pitches) > 1 and pitch_mean > 0 else 0
         intensities = []
         for segment in segments:
-            rms = np.mean(librosa.feature.rms(y=segment)[0])
-            intensities.extend(rms)
+            rms = np.mean(librosa.feature.rms(y=segment)[0]) if len(segment) > 0 else 0.0
+            intensities.append(float(rms))  # Fix: Use append instead of extend
         intensity_mean = np.mean(intensities) if intensities else 0
         intensity_std = np.std(intensities) if intensities else 0
-        shimmer = np.mean(np.abs(np.diff(intensities))) / intensity_mean if intensity_mean > 0 else 0
+        shimmer = np.mean(np.abs(np.diff(intensities))) / intensity_mean if len(intensities) > 1 and intensity_mean > 0 else 0
         anxiety_score = 0.6 * (pitch_std / pitch_mean) + 0.4 * (jitter + shimmer) if pitch_mean > 0 else 0
         confidence_score = 0.7 * (1 / (1 + intensity_std)) + 0.3 * (1 / (1 + filler_ratio))
         hesitation_score = filler_ratio + repetition_score
@@ -376,12 +383,12 @@ def analyze_interviewee_voice(audio_path: str, utterances: List[Dict]) -> Dict:
             'interpretation': {'anxiety_level': anxiety_level, 'confidence_level': confidence_level, 'fluency_level': fluency_level}
         }
     except Exception as e:
-        logger.error(f"Voice analysis failed: {str(e)}")
-        return {'error': str(e)}
+        logger.error(f"Voice analysis failed: {str(e)}", exc_info=True)
+        return {'error': f'Voice analysis incomplete due to audio processing issues: {str(e)}'}
 
 def generate_voice_interpretation(analysis: Dict) -> str:
     if 'error' in analysis:
-        return "Voice analysis unavailable due to processing limitations."
+        return f"Voice analysis unavailable: {analysis['error']}"
     interpretation_lines = [
         f"- Speaking Rate: {analysis['speaking_rate']} words/sec (Benchmark: 2.0-3.0 wps; affects clarity)",
         f"- Filler Words: {analysis['filler_ratio'] * 100:.1f}% (High usage reduces credibility)",
@@ -413,7 +420,7 @@ def generate_anxiety_confidence_chart(composite_scores: Dict, chart_buffer):
                     ha='center', color='black', fontweight='bold', fontsize=10)
         ax.grid(True, axis='y', linestyle='--', alpha=0.7)
         plt.tight_layout()
-        plt.savefig(chart_buffer, format='png', bbox_inches='tight', dpi=300)
+        plt.savefig(chart_buffer, format='png', bbox_inches='tight', dpi=100)
         plt.close(fig)
     except Exception as e:
         logger.error(f"Error generating chart: {str(e)}")
@@ -516,10 +523,10 @@ def create_pdf_report(analysis_data: Dict, output_path: str, gemini_report_text:
         story.append(Paragraph("Candidate Interview Analysis", h1))
         story.append(Paragraph(f"Generated: {time.strftime('%B %d, %Y')}", ParagraphStyle(name='Date', alignment=1, fontSize=8, textColor=colors.HexColor('#666666'), fontName='Helvetica')))
         story.append(Spacer(1, 0.3*inch))
-        acceptance_prob = analysis_data.get('acceptance_probability', 50.0)
+        acceptance_prob = float(np.mean([np.mean([np.mean([analysis_data['acceptance_probability'], 0.0])])]))  # Ensure float
         story.append(Paragraph("Hiring Suitability Snapshot", h2))
         prob_color = colors.HexColor('#2E7D32') if acceptance_prob >= 80 else (colors.HexColor('#F57C00') if acceptance_prob >= 60 else colors.HexColor('#D32F2F'))
-        story.append(Paragraph(f"Suitability Score: <font size=14 color='{prob_color.hexval()}'><b>{acceptance_prob:.2f}%</b></font>",
+        story.append(Paragraph(f"Suitability Score: <font size=14 color='{prob_color.hexval()}'>{acceptance_prob:.2f}%</font>", 
                              ParagraphStyle(name='Prob', fontSize=10, spaceAfter=8, alignment=1, fontName='Helvetica-Bold')))
         if acceptance_prob >= 80:
             story.append(Paragraph("<b>HR Verdict:</b> Outstanding candidate, recommended for immediate advancement.", body_text))
@@ -533,9 +540,9 @@ def create_pdf_report(analysis_data: Dict, output_path: str, gemini_report_text:
         participants = sorted([p for p in set(u['speaker'] for u in analysis_data['transcript']) if p != 'Unknown'])
         table_data = [
             ['Metric', 'Value'],
-            ['Interview Duration', f"{analysis_data['text_analysis']['total_duration']:.2f} seconds"],
+            ['Interview Duration', f"{analysis_data['text_analysis']['total_duration']:.1f} seconds"],
             ['Speaker Turns', f"{analysis_data['text_analysis']['speaker_turns']}"],
-            ['Participants', ', '.join(participants)],
+            ['Participants', f"{', '.join(participants)}"],
         ]
         table = Table(table_data, colWidths=[2.2*inch, 3.8*inch])
         table.setStyle(TableStyle([
@@ -547,7 +554,7 @@ def create_pdf_report(analysis_data: Dict, output_path: str, gemini_report_text:
             ('FONTSIZE', (0,0), (-1,-1), 8),
             ('BOTTOMPADDING', (0,0), (-1,0), 6),
             ('TOPPADDING', (0,0), (-1,0), 6),
-            ('BACKGROUND', (0,1), (-1,-1), colors.HexColor('#F5F6FA')),
+            ('BACKGROUND', (0,1), (-1,-1), colors.HexColor('#F5F6FA'),),
             ('GRID', (0,0), (-1,-1), 0.4, colors.HexColor('#DDE4EB')),
         ]))
         story.append(table)
@@ -558,41 +565,41 @@ def create_pdf_report(analysis_data: Dict, output_path: str, gemini_report_text:
         # Detailed Analysis
         story.append(Paragraph("Detailed Candidate Evaluation", h1))
         
-        # Communication and Vocal Dynamics
+        # Communication and Speech
         story.append(Paragraph("1. Communication & Vocal Dynamics", h2))
         voice_analysis = analysis_data.get('voice_analysis', {})
-        if voice_analysis and 'error' not in voice_analysis:
+        if voice_analysis' and 'error' not in voice_analysis:
             table_data = [
                 ['Metric', 'Value', 'HR Insight'],
                 ['Speaking Rate', f"{voice_analysis.get('speaking_rate', 0):.2f} words/sec", 'Benchmark: 2.0-3.0 wps; impacts clarity'],
-                ['Filler Words', f"{voice_analysis.get('filler_ratio', 0) * 100:.1f}%", 'High usage reduces credibility'],
+                ['Filler Words', f"{voice_analysis.get('filler_ratio', 0) * 100:.1f}%', 'High usage reduces credibility'],
                 ['Anxiety', voice_analysis.get('interpretation', {}).get('anxiety_level', 'N/A'), f"Score: {voice_analysis.get('composite_scores', {}).get('anxiety', 0):.3f}"],
                 ['Confidence', voice_analysis.get('interpretation', {}).get('confidence_level', 'N/A'), f"Score: {voice_analysis.get('composite_scores', {}).get('confidence', 0):.3f}"],
                 ['Fluency', voice_analysis.get('interpretation', {}).get('fluency_level', 'N/A'), 'Drives engagement'],
             ]
             table = Table(table_data, colWidths=[1.5*inch, 1.3*inch, 3.2*inch])
             table.setStyle(TableStyle([
-                ('BACKGROUND', (0,0), (-1,0), colors.HexColor('#0050BC')),
-                ('TEXTCOLOR', (0,0), (-1,0), colors.white),
+                ('BACKGROUND', (0,0), (-1,0)), colors.HexColor('#0050BC')),
+                ('TEXTCOLOR', (0,0), (-1,-0)), colors.white),
                 ('ALIGN', (0,0), (-1,-1), 'LEFT'),
                 ('VALIGN', (0,0), (-1,-1), 'MIDDLE'),
-                ('FONTNAME', (0,0), (-1,0), 'Helvetica-Bold'),
+                ('FONTNAME', (0,0), (-1,-0)), 'Helvetica-Bold'),
                 ('FONTSIZE', (0,0), (-1,-1), 8),
-                ('BOTTOMPADDING', (0,0), (-1,0), 6),
-                ('TOPPADDING', (0,0), (-1,0), 6),
-                ('BACKGROUND', (0,1), (-1,-1), colors.HexColor('#F5F6FA')),
-                ('GRID', (0,0), (-1,-1), 0.4, colors.HexColor('#DDE4EB')),
+                ('BOTTOMPADDING', (0,0), (-1,-0)), 6),
+                ('TOPPADDING', (0,0), (0,-1), 6),
+                ('BACKGROUND', (0,1), (-1,-1), colors.HexColor('#F5F6FA'))),
+                ('GRID', (0,0), (-1,-1), 0.4, colors.HexColor('#DDE4EB'))),
             ]))
             story.append(table)
             story.append(Spacer(1, 0.15*inch))
             chart_buffer = io.BytesIO()
-            generate_anxiety_confidence_chart(voice_analysis.get('composite_scores', {}), chart_buffer)
+            generate_anxiety_chart(voice_analysis.get('composite_scores', {}), chart_buffer)
             chart_buffer.seek(0)
             img = Image(chart_buffer, width=4.2*inch, height=2.8*inch)
             img.hAlign = 'CENTER'
             story.append(img)
         else:
-            story.append(Paragraph("Vocal analysis unavailable due to processing limitations.", body_text))
+            story.append(Paragraph("Voice analysis unavailable.", body_text))
         story.append(Spacer(1, 0.15*inch))
 
         # Parse Gemini Report
@@ -628,12 +635,12 @@ def create_pdf_report(analysis_data: Dict, output_path: str, gemini_report_text:
                 elif 'Recommendations' in section_title:
                     current_section = 'Recommendations'
                     current_subsection = None
-            elif line.startswith(('-', '*', '•')) and current_section:
-                clean_line = line.lstrip('-*• ').strip()
+            elif line.startswith('-') and current_section:
+                clean_line = line.lstrip('-').strip()
                 if not clean_line: continue
                 clean_line = re.sub(r'[()]', '', clean_line)
                 if current_section == 'Competency':
-                    if any(k in clean_line.lower() for k in ['leader', 'problem', 'commun', 'adapt', 'strength']):
+                    if any(k in clean_line.lower() for k in ['leader', 'leadership', 'problem', 'commun', 'adapt', 'strength']):
                         current_subsection = 'Strengths'
                     elif any(k in clean_line.lower() for k in ['improv', 'grow', 'depth']):
                         current_subsection = 'Growth Areas'
@@ -793,6 +800,6 @@ def process_interview(audio_url: str) -> Dict:
         if is_downloaded and local_audio_path and os.path.exists(local_audio_path):
             try:
                 os.remove(local_audio_path)
-                logger.info(f"Cleaned up temporary file: {local_audio_path}")
+                logger.info(f"Cleaned up temporary audio file: {local_audio_path}")
             except Exception as e:
                 logger.error(f"Failed to clean up local audio file {local_audio_path}: {str(e)}")