o huh*@sddlZddlZddlZddlmZddlZddlZddlmZm Z m Z m Z m Z m Z eeZddZdedeefdd Zd ed ed ed efddZdedefddZdS)N) BeautifulSoup)DATA_ATTR_SONG_IDDATA_ATTR_START DATA_ATTR_ENDDEFAULT_AUDIO_FORMATgenerate_audio_element_idgenerate_audio_filenamec CsVztjddgddddWdStjtfy*tdtdtdYdSw) z-Checks if yt-dlp is installed and executable.yt-dlpz --versionTcheckZcapture_outputtextz.yt-dlp command not found or failed to execute.z r zSuccessfully downloaded: zyt-dlp finished for z, but expected file 'z' not found. Check z for files starting with ''.z.*zFound potential match(es): FzFailed to download segment for z].z Command:  zStderr: zTyt-dlp command not found during execution. Please ensure it's installed and in PATH.z1An unexpected error occurred during download for z]: )rrospathjoinexistsrinfomakedirsrr rdebugr!globrrcmdstderrr Exception) rrrr% element_idZoutput_filename_baseoutput_filename_with_extoutput_filepathZ video_urlcommand_Zpotential_fileserrrdownload_audio_segmentBsZ    " $  "rAhtml_diraudio_output_dirc Cststdtj|ddttj|d}t dt |d|dt }|D]g}zHt |dd d  }| }Wd n1sEwYt|}tdt |d tj|d |D]}|d|d|df} || qbWq,ty} ztd|d| WYd } ~ q,d } ~ wwd} d} d} |D]F\}}}t|||}t|}tj||}tj|r| d7} td|qt||||}|rtj|r| d7} q| d7} q| d7} qt dt dt |t d| t d| t d| d S)aU Finds all HTML files in a directory, extracts audio tag data, and downloads the required audio segments into the specified `audio_output_dir`. Args: html_dir: Absolute path to the directory containing generated HTML files. audio_output_dir: Absolute path to the directory where audio files should be saved. z:yt-dlp is required but not found. Aborting audio download.Tr)z*.htmlzFound z HTML files in 'r.rzutf-8)encodingNz+ audio segments to potentially download in .rrrzError processing file z: rzSegment already exists: z--- Audio Download Summary ---zTotal unique segments found: zSuccessfully downloaded: zAlready existed: zFailed downloads: )rsysexitr0r5r7r1r2rr4lensetopenreadr$r6basenameaddr:rrrr3rA)rBrCZ html_filesZall_audio_segmentsZhtml_file_pathfcontentZ audio_tagsZtag_dataZ segment_idr@Zdownload_countZ fail_countZalready_exists_countrrrr;r<r=Zsuccessrrrprocess_html_filess\  $  "       rR)r0r rHZbs4rr7loggingmarkdown_extensionsrrrrrr getLogger__name__rrstrlistdictr$rrArRrrrrs    #Z