Add role detection, gap analysis, and UI optimization to resume analysis #25
Conversation
WalkthroughRefactors Changes
Sequence Diagram(s)sequenceDiagram
actor User
participant UI as Streamlit UI
participant App as app.py
participant LLM as LLM Service
participant Parser as JSON Extractor
User->>UI: Upload PDF + Select Language
UI->>App: Call analyzecv_pdf_withllm(text, language)
App->>LLM: Send crafted prompt with CV text
LLM-->>App: Return raw LLM output
App->>Parser: _extract_json_block(raw_output)
Parser-->>App: Extracted JSON string
App->>App: Parse & validate JSON -> Dict[str, Any]
alt Valid JSON
App-->>UI: Return parsed result
UI->>App: Call display_analysis_tabs(result)
App->>UI: Render display_summary_block()
App->>UI: Create tabs and render:
App->>UI: - Fit & Gaps
App->>UI: - Competencies
App->>UI: - Domains
App->>UI: - Insights
App->>UI: - Recommendations
else Invalid JSON
App-->>UI: Return None + error message
end
Estimated code review effort🎯 4 (Complex) | ⏱️ ~50 minutes
Poem
Pre-merge checks and finishing touches✅ Passed checks (2 passed)
✨ Finishing touches
🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 1
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (2)
levelup/app.py(4 hunks)levelup/prompts.py(4 hunks)
🧰 Additional context used
🧬 Code graph analysis (1)
levelup/app.py (1)
levelup/prompts.py (1)
get_resume_analysis_prompt(1-107)
| df = pd.DataFrame( | ||
| [ | ||
| {"Role": r.get("role", ""), "Score": r.get("score", "")} | ||
| for r in role_suit | ||
| ] | ||
| ) | ||
| st.dataframe(df, width="stretch") |
There was a problem hiding this comment.
Fix invalid st.dataframe width usage
Passing width="stretch" causes StreamlitAPIException because the width parameter only accepts an integer or None. Right now the app will crash whenever these DataFrames render. Switch to use_container_width=True (or drop the argument) everywhere this pattern appears.
- st.dataframe(df, width="stretch")
+ st.dataframe(df, use_container_width=True)
@@
- st.dataframe(pd.DataFrame(rows), width="stretch")
+ st.dataframe(pd.DataFrame(rows), use_container_width=True)
@@
- st.dataframe(df, width="stretch")
+ st.dataframe(df, use_container_width=True)
@@
- st.dataframe(df, width="stretch")
+ st.dataframe(df, use_container_width=True)Also applies to: 200-208, 232-244, 258-268
🤖 Prompt for AI Agents
In levelup/app.py around lines 185-191 (and also update the other affected
occurrences at 200-208, 232-244, 258-268), the st.dataframe calls pass
width="stretch" which raises StreamlitAPIException; remove the invalid width
argument and replace it with use_container_width=True (or omit the width
argument entirely) for each DataFrame rendering so the calls become
st.dataframe(df, use_container_width=True).
There was a problem hiding this comment.
Actionable comments posted: 1
♻️ Duplicate comments (4)
levelup/app.py (4)
175-181: Fix thewidth="stretch"parameter (duplicate issue).This
width="stretch"parameter onst.dataframewas already flagged in previous reviews and will cause aStreamlitAPIExceptionat runtime. This specific instance at line 181 was noted in the prior comment covering lines 175-181.Apply this diff:
- st.dataframe(df, width="stretch") + st.dataframe(df, use_container_width=True)
190-197: Fix thewidth="stretch"parameter (duplicate issue).Another instance of the
width="stretch"parameter that will crash at runtime. This was flagged in previous reviews covering lines 200-208.Apply this diff:
- st.dataframe(pd.DataFrame(rows), width="stretch") + st.dataframe(pd.DataFrame(rows), use_container_width=True)
222-233: Fix thewidth="stretch"parameter (duplicate issue).Yet another
width="stretch"that needs correction. This was flagged in the prior review covering lines 232-244.Apply this diff:
- st.dataframe(df, width="stretch") + st.dataframe(df, use_container_width=True)
248-258: Fix thewidth="stretch"parameter (duplicate issue).Final instance of the
width="stretch"error. This specific line 258 was in the range 258-268 flagged previously.Apply this diff:
- st.dataframe(df, width="stretch") + st.dataframe(df, use_container_width=True)
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (1)
levelup/app.py(4 hunks)
🧰 Additional context used
🧬 Code graph analysis (1)
levelup/app.py (1)
levelup/prompts.py (1)
get_resume_analysis_prompt(1-107)
🪛 GitHub Actions: CI
levelup/app.py
[error] 59-59: mypy: Return value expected. Found 1 error in 1 file (checked 13 source files).
🔇 Additional comments (8)
levelup/app.py (8)
24-33: LGTM: Improved text extraction with newline separators.The refactored PDF extraction now joins page text with newlines, which better preserves document structure for downstream analysis.
36-43: LGTM: Robust JSON extraction strategy.The two-stage fallback (fenced block → raw braces) handles common LLM output variations effectively.
69-71: LGTM: Useful defensive helper.The
_safe_dictutility provides safe nested dictionary access with appropriate fallback.
74-149: LGTM: Enhanced type safety and defensive data access.The updated display functions now have explicit type signatures and robust defaults for missing or malformed data, improving reliability when handling LLM responses.
263-271: LGTM: Well-structured insights layout.The 3:2 column ratio and string validation for benchmarking data provide a clean presentation.
274-281: LGTM: Clear recommendations display.Simple and effective presentation with appropriate fallback messaging.
284-304: LGTM: Clean tabbed UI orchestration.The modular tab structure makes the UI easy to extend and maintain, with clear separation between the summary block and tabbed sections.
307-331: LGTM: Intuitive main UI flow.The upload → configure → analyze → display flow is clear and user-friendly, with appropriate loading feedback.
There was a problem hiding this comment.
Actionable comments posted: 0
♻️ Duplicate comments (5)
levelup/app.py (5)
57-61: Fix the bare return statement (critical mypy error).Line 59 still has a bare
returnstatement, but the function signature requiresdict[str, Any] | None. This was flagged in a previous review but remains unfixed, causing the mypy pipeline failure.Apply this diff:
if not isinstance(data := json.loads(json_str), dict): st.error("Invalid JSON object.") - return + return NoneAdditionally, the
caston line 61 is redundant since you've already verifieddatais adictviaisinstanceon line 57:- return cast(dict[str, Any], data) + return data
172-181: Fix invalidwidth="stretch"parameter (critical Streamlit error).Line 181 uses
width="stretch", which causes aStreamlitAPIExceptionbecausest.dataframeonly accepts integers orNonefor thewidthparameter. This was flagged in a previous review but remains unfixed and will crash the app when role suitability data is displayed.Apply this diff:
- st.dataframe(df, width="stretch") + st.dataframe(df, use_container_width=True)
197-197: Fix invalidwidth="stretch"parameter (critical Streamlit error).Line 197 uses
width="stretch", causing aStreamlitAPIException. This was flagged in a previous review (lines 200-208 reference) but remains unfixed.Apply this diff:
- st.dataframe(pd.DataFrame(rows), width="stretch") + st.dataframe(pd.DataFrame(rows), use_container_width=True)
233-233: Fix invalidwidth="stretch"parameter (critical Streamlit error).Line 233 uses
width="stretch", causing aStreamlitAPIException. This was flagged in a previous review (lines 232-244 reference) but remains unfixed.Apply this diff:
- st.dataframe(df, width="stretch") + st.dataframe(df, use_container_width=True)
258-258: Fix invalidwidth="stretch"parameter (critical Streamlit error).Line 258 uses
width="stretch", causing aStreamlitAPIException. This was flagged in a previous review (lines 258-268 reference) but remains unfixed.Apply this diff:
- st.dataframe(df, width="stretch") + st.dataframe(df, use_container_width=True)
🧹 Nitpick comments (1)
levelup/app.py (1)
69-72: LGTM with optional enhancement suggestion.The helper correctly guards against missing or non-dict values. If you need flexibility in the default return value, consider adding a parameter:
-def _safe_dict(obj: dict[str, Any], key: str) -> dict[str, Any]: +def _safe_dict(obj: dict[str, Any], key: str, default: dict[str, Any] | None = None) -> dict[str, Any]: + if default is None: + default = {} val = obj.get(key, {}) - return val if isinstance(val, dict) else {} + return val if isinstance(val, dict) else default
📜 Review details
Configuration used: CodeRabbit UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (1)
levelup/app.py(4 hunks)
🧰 Additional context used
🧬 Code graph analysis (1)
levelup/app.py (1)
levelup/prompts.py (1)
get_resume_analysis_prompt(1-107)
🔇 Additional comments (12)
levelup/app.py (12)
4-4: LGTM!The new imports are appropriate for the added functionality:
castfor type narrowing andpandasfor DataFrame rendering. The type ignore comment on pandas is correct since it lacks inline type stubs.Also applies to: 7-7
27-30: LGTM!The refactored per-page text extraction with list accumulation and newline joining is more efficient and preserves page structure better than string concatenation.
36-43: LGTM!The JSON extraction logic correctly handles both fenced code blocks (with optional
jsontag) and bare JSON objects. The two-tier matching strategy (fenced first, then bare braces) provides good robustness for various LLM output formats.
79-92: LGTM!The display functions demonstrate good defensive programming:
- Consistent use of
.get()with sensible defaults- The
or []pattern safely handlesNonevalues from JSON- Line 126's type check prevents displaying non-string benchmarking data
Also applies to: 95-109, 117-120, 123-126, 129-139
152-171: LGTM!The summary block layout is well-structured:
- Clean three-column metrics display
- Good use of
_safe_dictfor safe access- Conditional markdown rendering for strengths/areas avoids empty lists
184-196: LGTM with good defensive programming.The two-column layout clearly separates missing skills from mismatched experience. Line 193's
str()cast is a good defensive measure to handle non-string priority values.Also applies to: 198-208
210-232: LGTM with excellent defensive programming.The competencies tab demonstrates robust error handling:
- Lines 215-219: Safe float conversion with try-except fallback
- Line 221: Sorting by score improves readability
- Line 235: Helpful info message when no data is available
Also applies to: 234-236
238-257: LGTM with robust error handling.The domains tab handles sorting failures gracefully (lines 243-247), falling back to the original order if score extraction fails. The info message on line 260 provides clear feedback when data is absent.
Also applies to: 259-261
263-272: LGTM!The insights tab has a clean two-column layout with appropriate proportions. Line 271's defensive check ensures only valid, non-empty strings are displayed.
274-282: LGTM!The recommendations tab has a clean, straightforward implementation with appropriate fallback messaging.
284-304: LGTM!The tab orchestration is well-structured with clear separation of concerns. Displaying the summary block before the tabs provides users with a high-level overview before diving into details.
309-331: LGTM!The main application flow is clean and intuitive:
- Sequential steps guide the user through the process
- Spinner provides feedback during analysis
- Conditional result display ensures results are only shown when available
Summary by CodeRabbit
New Features
Improvements