o 4i<@s@dZddlZddlZddlmZGdddZeZddZdS)zu Course Statistics Utility Analyzes the course outline file (output/{hashid}.md) to provide module and lesson counts N)Pathc@sneZdZdddZddZddZdd Zd d Zd d ZddZ ddZ ddZ ddZ ddZ ddZdS) CourseStatsNcCsH|durtjtjt}tj||_td|jdS||_dS)z!Initialize with project root pathNu!📁 Auto-detected project root: )ospathdirnameabspath__file__ project_rootprint)selfr current_dirr u/home/azureuser/microlearn/backend/user_journey_with_openai/agentic_workflow/src/user_journey_service/course_stats.py__init__ s  zCourseStats.__init__c stj|jd|d}td|tdtjtj|tdtj|tj|sAtd||dd|d Sztd t|d d d  }|}Wdn1s\wY| ||t t s|tdt |dddWSgd}fdd|D}|rtd|tdt |D]%}|dkrdd<q|dkrdd<q|dkrdd<q|dkrgd<qWSty}z"tdt|ddl} | |ddt|dWYd}~Sd}~ww) z Analyze the course outline file and return statistics Args: hashid: The hashid of the course Returns: dict: Course statistics including module count, lesson counts, etc. output.mdu$🔍 Looking for course outline at: u📁 Output directory exists: u📄 File exists: u❌ File does not exist at: Fz=Course outline file not found. Please run /run-journey first.)hashidexistserrorsearched_locationu📖 Opening file...rzutf-8)encodingNu❌ Stats is not a dictionary: Tz&Parser returned invalid data structure)rrr) course_title module_count total_lessonsmodulescsg|]}|vr|qSr r ).0keystatsr r Csz0CourseStats.get_course_stats..u$❌ Missing required keys in stats: u📊 Stats keys found: rUnknown Courserrrru❌ Error parsing file: zError parsing course outline: )rrjoinr r rropenread_parse_course_outline isinstancedicttypelistkeys Exceptionstr traceback print_exc) r r outline_pathfcontent required_keys missing_keysrer-r rrget_course_statssb         zCourseStats.get_course_statsc CsRtdtdtdtd|tj}|r|dnd}td|d}t||tj}t|}td|d g}d } t|dd D]\} \} } t| } | } td | d | | || } | | }| | }| | }d}t|| tj}t|}| |7} td|dg}|D]2\}}| }td|d ||| |}||}||}||}||||||dq|| | |||||dqD| }tdtdtdtd|td|td| td| td|td|d||| || ||dd|dd|d d!|d d"|d d#d$S)%z6Parse the markdown content to extract course structurez2==================================================u📝 Parsing course outline...z^#\s+\*\*(.+?)\*\*r!u📚 Course Title: z'^##\s+\*\*Module\s+(\d+):\s*([^*]+)\*\*u 📊 Found z modulesr)startu 📦 Module z: z,###\s+\*\*Lesson\s+(\d+\.\d+):\s*([^*]+)\*\*u 📖 Found  lessonsu • Lesson ) lesson_number lesson_titledurationtopicsoutcome) module_number module_titlefocusr=r; lesson_countlessonsz3 ==================================================u 📊 SUMMARYzCourse: zTotal Modules: zTotal Lessons: zEstimated Content Files: zEstimated Audio Files: T) content_filescontent_audio_filestotal_audio_filesz/output/rz /content//z/updated_content/z/audio/lessons/)course_outlinecontent_directoryupdated_content_directoryaudio_lessons_directory)rrrrrrestimated_output file_paths)r research MULTILINEgroupfindalllen enumerateintstrip_extract_module_section_extract_module_focus_extract_module_outcome_extract_module_duration_extract_lesson_section_extract_lesson_duration_extract_lesson_topics_extract_lesson_outcomeappend)r r1rcourse_title_matchrmodule_patternrr modules_listr module_idx module_numr?module_section module_focusmodule_outcomemodule_durationlesson_patternrBrAlesson_details lesson_numr:lesson_sectionlesson_duration lesson_topicslesson_outcometotal_content_audior r rr%^s                z!CourseStats._parse_course_outlinec Csd|d}t||tj}|sdS|}d|dd}t|||ddtj}|r<|d|}|||S||dS)z=Extract the section of content belonging to a specific modulez^##\s+\*\*Module\s+:r6N)rMrNrOr7) r r1rc start_pattern start_match start_posnext_module_pattern next_matchend_posr r rrVs   z#CourseStats._extract_module_sectioncC&td|tj}|r|dSdS)z(Extract module focus from module sectionz.### \*\*Module Focus\*\*\s*\n(.*?)(?=\n###|\Z)r6rqrMrNDOTALLrPrU)r section focus_matchr r rrWz!CourseStats._extract_module_focuscCrx)z*Extract module outcome from module sectionz0### \*\*Module Outcome\*\*\s*\n(.*?)(?=\n###|\Z)r6rqry)r r{ outcome_matchr r rrXr}z#CourseStats._extract_module_outcomecCrx)z+Extract module duration from module sectionz7### \*\*Total Module Duration\*\*\s*\n(.*?)(?=\n###|\Z)r6rqry)r r{duration_matchr r rrYr}z$CourseStats._extract_module_durationcCs4dt|d}t||tj}|r|dSdS)z5Extract a specific lesson section from module sectionz### \*\*Lesson z$:.*?\*\*(.*?)(?=\n### \*\*Lesson|\Z)r6rq)rMescaperNrzrP)r rdrjpatternmatchr r rrZsz#CourseStats._extract_lesson_sectioncCrx)z+Extract lesson duration from lesson sectionz,#### \*\*Duration\*\*\s*\n(.*?)(?=\n####|\Z)r6rqry)r rkrr r rr[r}z$CourseStats._extract_lesson_durationcCsRtd|tj}|r'|d}td|tj}|s%dd|dD}|SgS)z)Extract lesson topics from lesson sectionz2#### \*\*Topics Covered\*\*\s*\n(.*?)(?=\n####|\Z)r6u[-•]\s*(.*?)(?=\n[-•]|\Z)cSsg|] }|r|qSr )rU)rtr r rr sz6CourseStats._extract_lesson_topics.. )rMrNrzrPrUrQsplit)r rk topics_match topics_textr<r r rr\sz"CourseStats._extract_lesson_topicscCrx)z*Extract lesson outcome from lesson sectionz+#### \*\*Outcome\*\*\s*\n(.*?)(?=\n####|\Z)r6rqry)r rkr~r r rr]r}z#CourseStats._extract_lesson_outcomec Csh||}|r t|tstd||dddddS|dr8td|d|d|dd d|d d d S|d dsJ|dd d|d d dSzC|d|ddd|ddd|ddd|dd|dd|didd|diddddd|dgDdWSty}ztd ||dd!t|d"d#dWYd$}~Sd$}~ww)%zGGet course statistics in a nicely formatted dictionary for API responseu❌ Invalid stats returned: FzFailed to parse course outlinezXCourse outline not found or invalid format. Please ensure /run-journey was called first.z4Jobs are still queued but course outline is missing.)rrrwarningnoteru❌ Error in stats: zFCourse outline not found. Please ensure /run-journey was called first.runknown)rrrrrrr)rrrrrTrr!zCourse contains rrz modules with rr8rKrCrE)rC audio_filescSs,g|]}|d|d|dddqS)r>r?rAr)r>titlerA)get)rmr r rr <s z3CourseStats.get_formatted_stats..r)rrrsummaryrrestimated_filesru❌ Error formatting stats: zError formatting course stats: z7Course outline exists but could not be parsed properly.z8Jobs are still queued but course preview is unavailable.N)r5r&r'r rr+r,)r rrr4r r rget_formatted_stats sb            zCourseStats.get_formatted_stats)N)__name__ __module__ __qualname__rr5r%rVrWrXrYrZr[r\r]rr r r rr s  Hk rcCs t|S)z(Convenience function to get course stats) course_statsr)rr r rr5Ts r5)__doc__rrMpathlibrrrr5r r r rs J