Update SCRAPE_MEET.SH
This commit is contained in:
parent
f694de0674
commit
3a76f2f5af
207
SCRAPE_MEET.SH
207
SCRAPE_MEET.SH
@ -108,76 +108,102 @@ clear_agenda_url() {
|
||||
ADDENDUM_PDF_URL=""
|
||||
}
|
||||
|
||||
download_helper() {
|
||||
local url="$1"
|
||||
local out="$2"
|
||||
local code
|
||||
|
||||
code=$(curl -L -k -A "$WGET_UA" -sS -w "%{http_code}" --retry 3 --retry-delay 2 -z "$out" -o "$out" "$url")
|
||||
case "$code" in
|
||||
200)
|
||||
echo "Downloaded."
|
||||
;;
|
||||
304)
|
||||
echo "Already exists! Skipping."
|
||||
;;
|
||||
*)
|
||||
echo "FAILED! $code: $out | $url" >&2
|
||||
return 1
|
||||
;;
|
||||
esac
|
||||
}
|
||||
|
||||
download_agendas() {
|
||||
if [[ $AGENDA_REVISE_PDF_URL != "" ]] || [[ $AGENDA_PDF_URL != "" ]]; then
|
||||
if [[ $AGENDA_REVISE_PDF_URL != "" ]]; then
|
||||
local outdir="$1"
|
||||
|
||||
if [[ -n $AGENDA_REVISE_PDF_URL ]]; then
|
||||
echo "Saving revised agenda as PDF..."
|
||||
wget --no-check-certificate --user-agent="$WGET_UA" "$AGENDA_REVISE_PDF_URL" -O "$1/Agenda_Revised.pdf" -N -q #--show-progress
|
||||
download_helper "$AGENDA_REVISE_PDF_URL" "$outdir/Agenda_Revised.pdf"
|
||||
fi
|
||||
if [[ $AGENDA_PDF_URL != "" ]]; then
|
||||
|
||||
if [[ -n $AGENDA_PDF_URL ]]; then
|
||||
echo "Saving regular agenda as PDF..."
|
||||
wget --no-check-certificate --user-agent="$WGET_UA" "$AGENDA_PDF_URL" -O "$1/Agenda.pdf" -N -q #--show-progress
|
||||
download_helper "$AGENDA_PDF_URL" "$outdir/Agenda.pdf"
|
||||
fi
|
||||
elif [[ $AGENDA_REVISE_HTML_URL != "" ]] || [[ $AGENDA_HTML_URL != "" ]]; then
|
||||
if [[ $AGENDA_REVISE_HTML_URL != "" ]]; then
|
||||
|
||||
if [[ -z $AGENDA_REVISE_PDF_URL && -n $AGENDA_REVISE_HTML_URL ]]; then
|
||||
echo "Saving revised agenda as HTML... (no PDF found!)"
|
||||
wget --no-check-certificate --user-agent="$WGET_UA" "$AGENDA_REVISE_HTML_URL" -O "$1/Agenda_Revised.html" -N -q #--show-progress
|
||||
download_helper "$AGENDA_REVISE_HTML_URL" "$outdir/Agenda_Revised.html"
|
||||
fi
|
||||
if [[ $AGENDA_HTML_URL != "" ]]; then
|
||||
|
||||
if [[ -z $AGENDA_PDF_URL && -n $AGENDA_HTML_URL ]]; then
|
||||
echo "Saving regular agenda as HTML... (no PDF found!)"
|
||||
wget --no-check-certificate --user-agent="$WGET_UA" "$AGENDA_HTML_URL" -O "$1/Agenda.html" -N -q #--show-progress
|
||||
download_helper "$AGENDA_HTML_URL" "$outdir/Agenda.html"
|
||||
fi
|
||||
elif [[ $AGENDA_FULL_PDF_URL != "" ]] || [[ $AGENDA_FULL_HTML_URL != "" ]]; then
|
||||
if [[ $AGENDA_FULL_PDF_URL != "" ]]; then
|
||||
|
||||
if [[ -n $AGENDA_FULL_PDF_URL ]]; then
|
||||
echo "Saving full package agenda as PDF... (no HTML found!)"
|
||||
wget --no-check-certificate --user-agent="$WGET_UA" "$AGENDA_FULL_PDF_URL" -O "$1/Agenda_FullPackage.pdf" -N -q #--show-progress
|
||||
download_helper "$AGENDA_FULL_PDF_URL" "$outdir/Agenda_FullPackage.pdf"
|
||||
fi
|
||||
if [[ $AGENDA_FULL_HTML_URL != "" ]]; then
|
||||
|
||||
if [[ -z $AGENDA_FULL_PDF_URL && -n $AGENDA_FULL_HTML_URL ]]; then
|
||||
echo "Saving full package agenda as HTML... (no PDF found!)"
|
||||
wget --no-check-certificate --user-agent="$WGET_UA" "$AGENDA_FULL_HTML_URL" -O "$1/Agenda_FullPackage.html" -N -q #--show-progress
|
||||
download_helper "$AGENDA_FULL_HTML_URL" "$outdir/Agenda_FullPackage.html"
|
||||
fi
|
||||
elif [[ $AGENDA_POST_PDF_URL != "" ]] || [[ $AGENDA_POST_HTML_URL != "" ]]; then
|
||||
if [[ $AGENDA_POST_PDF_URL != "" ]]; then
|
||||
echo "Saving post agenda as HTML... (no HTML found!)"
|
||||
wget --no-check-certificate --user-agent="$WGET_UA" "$AGENDA_POST_PDF_URL" -O "$1/Agenda_Post.pdf" -N -q #--show-progress
|
||||
|
||||
if [[ -n $AGENDA_POST_PDF_URL ]]; then
|
||||
echo "Saving post agenda as PDF..."
|
||||
download_helper "$AGENDA_POST_PDF_URL" "$outdir/Agenda_Post.pdf"
|
||||
fi
|
||||
if [[ $AGENDA_POST_HTML_URL != "" ]]; then
|
||||
|
||||
if [[ -z $AGENDA_POST_PDF_URL && -n $AGENDA_POST_HTML_URL ]]; then
|
||||
echo "Saving post agenda as HTML... (no PDF found!)"
|
||||
wget --no-check-certificate --user-agent="$WGET_UA" "$AGENDA_POST_HTML_URL" -O "$1/Agenda_Post.html" -N -q #--show-progress
|
||||
fi
|
||||
download_helper "$AGENDA_POST_HTML_URL" "$outdir/Agenda_Post.html"
|
||||
fi
|
||||
|
||||
if [[ $MINUTES_ATTACH_PDF_URL != "" ]] || [[ $MINUTES_PDF_URL != "" ]]; then
|
||||
if [[ $MINUTES_ATTACH_PDF_URL != "" ]]; then
|
||||
if [[ -n $MINUTES_ATTACH_PDF_URL ]]; then
|
||||
echo "Saving minutes with attachments as PDF..."
|
||||
wget --no-check-certificate --user-agent="$WGET_UA" "$MINUTES_ATTACH_PDF_URL" -O "$1/Minutes_With_Attachments.pdf" -N -q #--show-progress
|
||||
fi
|
||||
if [[ $MINUTES_PDF_URL != "" ]]; then
|
||||
echo "Saving minutes as PDF..."
|
||||
wget --no-check-certificate --user-agent="$WGET_UA" "$MINUTES_PDF_URL" -O "$1/Minutes.pdf" -N -q #--show-progress
|
||||
fi
|
||||
else
|
||||
if [[ $MINUTES_HTML_URL != "" ]]; then
|
||||
echo "Saving minutes as HTML... (no PDF found!)"
|
||||
wget --no-check-certificate --user-agent="$WGET_UA" "$MINUTES_HTML_URL" -O "$1/Minutes.html" -N -q #--show-progress
|
||||
fi
|
||||
download_helper "$MINUTES_ATTACH_PDF_URL" "$outdir/Minutes_With_Attachments.pdf"
|
||||
fi
|
||||
|
||||
if [[ $AGENDA_COVER_PDF_URL != "" ]]; then
|
||||
if [[ -n $MINUTES_PDF_URL ]]; then
|
||||
echo "Saving minutes as PDF..."
|
||||
download_helper "$MINUTES_PDF_URL" "$outdir/Minutes.pdf"
|
||||
fi
|
||||
|
||||
if [[ -z $MINUTES_ATTACH_PDF_URL && -z $MINUTES_PDF_URL && -n $MINUTES_HTML_URL ]]; then
|
||||
echo "Saving minutes as HTML... (no PDF found!)"
|
||||
download_helper "$MINUTES_HTML_URL" "$outdir/Minutes.html"
|
||||
fi
|
||||
|
||||
if [[ -n $AGENDA_COVER_PDF_URL ]]; then
|
||||
echo "Saving cover agenda as PDF... (no HTML found!)"
|
||||
wget --no-check-certificate --user-agent="$WGET_UA" "$AGENDA_COVER_PDF_URL" -O "$1/Agenda_Cover.pdf" -N -q #--show-progress
|
||||
download_helper "$AGENDA_COVER_PDF_URL" "$outdir/Agenda_Cover.pdf"
|
||||
fi
|
||||
if [[ $AGENDA_COVER_HTML_URL != "" ]]; then
|
||||
|
||||
if [[ -z $AGENDA_COVER_PDF_URL && -n $AGENDA_COVER_HTML_URL ]]; then
|
||||
echo "Saving cover agenda as HTML... (no PDF found!)"
|
||||
wget --no-check-certificate --user-agent="$WGET_UA" "$AGENDA_COVER_HTML_URL" -O "$1/Agenda_Cover.html" -N -q #--show-progress
|
||||
download_helper "$AGENDA_COVER_HTML_URL" "$outdir/Agenda_Cover.html"
|
||||
fi
|
||||
if [[ $ADDENDUM_PDF_URL != "" ]]; then
|
||||
|
||||
if [[ -n $ADDENDUM_PDF_URL ]]; then
|
||||
echo "Saving addendum as PDF... (no HTML found!)"
|
||||
wget --no-check-certificate --user-agent="$WGET_UA" "$ADDENDUM_PDF_URL" -O "$1/Addendum.pdf" -N -q #--show-progress
|
||||
download_helper "$ADDENDUM_PDF_URL" "$outdir/Addendum.pdf"
|
||||
fi
|
||||
if [[ $ADDENDUM_HTML_URL != "" ]]; then
|
||||
|
||||
if [[ -z $ADDENDUM_PDF_URL && -n $ADDENDUM_HTML_URL ]]; then
|
||||
echo "Saving addendum as HTML... (no PDF found!)"
|
||||
wget --no-check-certificate --user-agent="$WGET_UA" "$ADDENDUM_HTML_URL" -O "$1/Addendum.html" -N -q #--show-progress
|
||||
download_helper "$ADDENDUM_HTML_URL" "$outdir/Addendum.html"
|
||||
fi
|
||||
}
|
||||
|
||||
@ -196,9 +222,9 @@ ADDENDUM_HTML="./tmp/addendum.html"
|
||||
|
||||
current_year=$(date +%Y)
|
||||
current_month=$(date +%m)
|
||||
current_day=$(date +%d)00
|
||||
current_day=$(date +%d)
|
||||
|
||||
SUPPORT_PAST=1
|
||||
SUPPORT_PAST=""
|
||||
|
||||
if [ -d "$TEMP_DIR" ]; then
|
||||
rm -r $TEMP_DIR
|
||||
@ -209,23 +235,23 @@ rm -f $AGENDA_HTML
|
||||
|
||||
mkdir $TEMP_DIR
|
||||
|
||||
while read -r INDEX_URL_PRE CITY_ARCHIVE_NAME_PRE CALENDAR_NAME_PRE; do
|
||||
INDEX_URL=$(echo $INDEX_URL_PRE | sed 's/\"//g' | sed 's/,//g')
|
||||
CITY_ARCHIVE_NAME=$(echo $CITY_ARCHIVE_NAME_PRE | sed 's/\"//g' | sed 's/\,//g')
|
||||
CALENDAR_NAME=$(echo $CALENDAR_NAME_PRE | sed 's/\"//g' | sed 's/\,//g')
|
||||
while IFS="," read -r INDEX_URL_PRE CITY_ARCHIVE_NAME_PRE CALENDAR_NAME_PRE; do
|
||||
INDEX_URL=$(echo "$INDEX_URL_PRE" | sed 's/\"//g' | sed 's/,//g' | sed 's/^[[:blank:]]*//;s/[[:blank:]]*$//')
|
||||
CITY_ARCHIVE_NAME=$(echo "$CITY_ARCHIVE_NAME_PRE" | sed 's/\"//g' | sed 's/\,//g' | sed 's/^[[:blank:]]*//;s/[[:blank:]]*$//')
|
||||
CALENDAR_NAME=$(echo "$CALENDAR_NAME_PRE" | sed 's/\"//g' | sed 's/\,//g' | sed 's/^[[:blank:]]*//;s/[[:blank:]]*$//')
|
||||
|
||||
INDEX_END=0
|
||||
while (( ! INDEX_END )); do
|
||||
INDEX_END="FALSE"
|
||||
while [[ $INDEX_END == "FALSE" ]]; do
|
||||
echo "SCRAPE_ESCRIBE: Downloading eScribe index..."
|
||||
wget --no-check-certificate --user-agent="$WGET_UA" $INDEX_URL -O $INDEX_PAGE --show-progress
|
||||
wget --no-check-certificate --user-agent="$WGET_UA" $INDEX_URL -O $INDEX_PAGE --no-hsts --show-progress
|
||||
if [ $? -ne 8 ]; then
|
||||
FOUNDLIST=0
|
||||
FOUNDLIST="FALSE"
|
||||
while IFS= read -r LINE; do
|
||||
if (( FOUNDLIST )); then
|
||||
if [[ "TRUE" == $FOUNDLIST ]]; then
|
||||
GREPENDLIST=$(echo $LINE | grep '<option ')
|
||||
if [[ "$GREPENDLIST" == "" ]]; then
|
||||
echo "SCRAPE_ESCRIBE: End of list."
|
||||
INDEX_END=1
|
||||
INDEX_END="TRUE"
|
||||
break
|
||||
else
|
||||
MEETING_NAME=$(echo $LINE | sed 's/.*<option[^>]*>\([^<]*\)<[\/:-]option>.*/\1/g')
|
||||
@ -312,11 +338,11 @@ while read -r INDEX_URL_PRE CITY_ARCHIVE_NAME_PRE CALENDAR_NAME_PRE; do
|
||||
else
|
||||
echo "Dates are in the past!"
|
||||
echo "DATE : $MEETING_YEAR/$MEETING_MONTH/$MEETING_DAY"
|
||||
INPAST=1
|
||||
INPAST="TRUE"
|
||||
fi
|
||||
|
||||
# I think "break" broke when I did nested loops. idk I'm too drunk for this.
|
||||
if (( INPAST )) && (( ! SUPPORT_PAST )); then
|
||||
if [[ "$INPAST" == "TRUE" ]] && [[ "$SUPPORT_PAST" != "TRUE" ]]; then
|
||||
echo "Abort."
|
||||
break
|
||||
fi
|
||||
@ -325,30 +351,37 @@ while read -r INDEX_URL_PRE CITY_ARCHIVE_NAME_PRE CALENDAR_NAME_PRE; do
|
||||
#cat "${TEMP_DIR}escribe_short.json" | jq --argjson i "$i" '.VideoLink.[].HasVideo'
|
||||
VIDEOURL="$INDEX_URL$(cat "${TEMP_DIR}escribe_short.json" | jq --argjson i "$i" '.VideoLink.[].Url' | sed 's/\"//g')"
|
||||
|
||||
ERROR=0
|
||||
ADDENDUM_ERROR=0
|
||||
ERROR="FALSE"
|
||||
ADDENDUM_ERROR="FALSE"
|
||||
echo "Downloading agenda HTML..."
|
||||
if [[ $AGENDA_REVISE_HTML_URL != "" ]]; then
|
||||
wget --no-check-certificate --user-agent="$WGET_UA" "$AGENDA_REVISE_HTML_URL" -O $AGENDA_HTML -q #--show-progress
|
||||
elif [[ $AGENDA_HTML_URL != "" ]]; then
|
||||
wget --no-check-certificate --user-agent="$WGET_UA" "$AGENDA_HTML_URL" -O $AGENDA_HTML -q #--show-progress
|
||||
elif [[ $AGENDA_FULL_HTML_URL != "" ]]; then
|
||||
wget --no-check-certificate --user-agent="$WGET_UA" "$AGENDA_FULL_HTML_URL" -O $AGENDA_HTML -q #--show-progress
|
||||
elif [[ $AGENDA_POST_HTML_URL != "" ]]; then
|
||||
wget --no-check-certificate --user-agent="$WGET_UA" "$AGENDA_POST_HTML_URL" -O $AGENDA_HTML -q #--show-progress
|
||||
elif [[ $AGENDA_COVER_HTML_URL != "" ]]; then
|
||||
wget --no-check-certificate --user-agent="$WGET_UA" "$AGENDA_COVER_HTML_URL" -O $AGENDA_HTML -q #--show-progress
|
||||
|
||||
if [[ -n $AGENDA_REVISE_HTML_URL ]]; then
|
||||
download_helper "$AGENDA_REVISE_HTML_URL" "$AGENDA_HTML"
|
||||
|
||||
elif [[ -n $AGENDA_HTML_URL ]]; then
|
||||
download_helper "$AGENDA_HTML_URL" "$AGENDA_HTML"
|
||||
|
||||
elif [[ -n $AGENDA_FULL_HTML_URL ]]; then
|
||||
download_helper "$AGENDA_FULL_HTML_URL" "$AGENDA_HTML"
|
||||
|
||||
elif [[ -n $AGENDA_POST_HTML_URL ]]; then
|
||||
download_helper "$AGENDA_POST_HTML_URL" "$AGENDA_HTML"
|
||||
|
||||
elif [[ -n $AGENDA_COVER_HTML_URL ]]; then
|
||||
download_helper "$AGENDA_COVER_HTML_URL" "$AGENDA_HTML"
|
||||
|
||||
else
|
||||
ERROR=1
|
||||
ERROR="TRUE"
|
||||
fi
|
||||
|
||||
if [[ $ADDENDUM_HTML_URL != "" ]]; then
|
||||
wget --no-check-certificate --user-agent="$WGET_UA" "$ADDENDUM_HTML_URL" -O $ADDENDUM_HTML -q #--show-progress
|
||||
|
||||
if [[ -n $ADDENDUM_HTML_URL ]]; then
|
||||
download_helper "$ADDENDUM_HTML_URL" "$ADDENDUM_HTML"
|
||||
else
|
||||
ADDENDUM_ERROR=1
|
||||
ADDENDUM_ERROR="TRUE"
|
||||
fi
|
||||
|
||||
if (( ! ERROR )); then
|
||||
if [[ "$ERROR" == "FALSE" ]]; then
|
||||
|
||||
mkdir "./$CITY_ARCHIVE_NAME"
|
||||
mkdir "./$CITY_ARCHIVE_NAME/Meetings"
|
||||
@ -373,20 +406,21 @@ while read -r INDEX_URL_PRE CITY_ARCHIVE_NAME_PRE CALENDAR_NAME_PRE; do
|
||||
fi
|
||||
|
||||
# Get attachment links
|
||||
cat $AGENDA_HTML | grep "AgendaItemAttachment AgendaItemAttachmentNotSelected" | sed 's/href=.filestream\.ashx/\nhref="filestream\.ashx/g' | grep 'filestream.ashx' | sed 's/. data-toggle/\" data-toggle/p' | sed 's/href=.\([^/]*\)".*/\1/p' | awk '!x[$0]++' > "./tmp/attachment_urls"
|
||||
cat $AGENDA_HTML | grep "AgendaItemAttachment AgendaItemAttachmentNotSelected" | sed 's/^.*AgendaHeaderTitle/AgendaHeaderTitle/' | sed 's/href=.[Ff]ile[Ss]tream\.ashx/\nhref="filestream\.ashx/g' | grep -i 'filestream.ashx' | sed 's/. data-toggle/\" data-toggle/p' | sed 's/href=.\([^"]*\)".*/\1/p' | awk '!x[$0]++' > "./tmp/attachment_urls"
|
||||
# Get attachment names
|
||||
cat $AGENDA_HTML | grep "AgendaItemAttachment AgendaItemAttachmentNotSelected" | sed 's/data-original-title=./\ndata-original-title='\''/g' | grep 'data-original-title' | sed 's/data-original-title=.//p' | sed 's/.pdf['\'':"].*/.pdf/g' | awk '!x[$0]++' > "./tmp/attachment_names"
|
||||
if (( ! ADDENDUM_ERROR )); then
|
||||
cat $AGENDA_HTML | grep "AgendaItemAttachment AgendaItemAttachmentNotSelected" | sed -E "s/data-original-title=['\"]/\\ndata-original-title='/g" | grep 'data-original-title' | sed 's/data-original-title=.//p' | sed 's/.pdf['\'':"].*/.pdf/g' | awk '!x[$0]++' > "./tmp/attachment_names"
|
||||
if [[ "$ADDENDUM_ERROR" == "FALSE" ]]; then
|
||||
# Get attachment links
|
||||
cat $ADDENDUM_HTML | grep "AgendaItemAttachment AgendaItemAttachmentNotSelected" | sed 's/href=.filestream\.ashx/\nhref="filestream\.ashx/g' | grep 'filestream.ashx' | sed 's/. data-toggle/\" data-toggle/p' | sed 's/href=.\([^/]*\)".*/\1/p' | awk '!x[$0]++' > "./tmp/attachment_urls"
|
||||
cat $ADDENDUM_HTML | grep "AgendaItemAttachment AgendaItemAttachmentNotSelected" | sed 's/^.*AgendaHeaderTitle/AgendaHeaderTitle/' | sed 's/href=.[Ff]ile[Ss]tream\.ashx/\nhref="filestream\.ashx/g' | grep -i 'filestream.ashx' | sed 's/. data-toggle/\" data-toggle/p' | sed 's/href=.\([^"]*\)".*/\1/p' | awk '!x[$0]++' > "./tmp/attachment_urls"
|
||||
# Get attachment names
|
||||
cat $ADDENDUM_HTML | grep "AgendaItemAttachment AgendaItemAttachmentNotSelected" | sed 's/data-original-title=./\ndata-original-title='\''/g' | grep 'data-original-title' | sed 's/data-original-title=.//p' | sed 's/.pdf['\'':"].*/.pdf/g' | awk '!x[$0]++' > "./tmp/attachment_names"
|
||||
cat $ADDENDUM_HTML | grep "AgendaItemAttachment AgendaItemAttachmentNotSelected" | sed -E "s/data-original-title=['\"]/\\ndata-original-title='/g" | grep 'data-original-title' | sed 's/data-original-title=.//p' | sed 's/.pdf['\'':"].*/.pdf/g' | awk '!x[$0]++' > "./tmp/attachment_names"
|
||||
fi
|
||||
# Download attachment and use the name grabbed above
|
||||
echo "Found the following agenda attachments:"
|
||||
while IFS= read -r LINEA1 && IFS= read -r LINEA2 <&3; do
|
||||
echo "- $LINEA2"
|
||||
wget --no-check-certificate --user-agent="$WGET_UA" "https://pub-london.escribemeetings.com/$LINEA1" -O "$MEETING_DIR/Attachments/$LINEA2" -N -q #--show-progress
|
||||
echo "- $LINEA2 / $LINEA1"
|
||||
download_helper "$INDEX_URL$LINEA1" "$MEETING_DIR/Attachments/$LINEA2"
|
||||
# [ ! -s "$MEETING_DIR/Attachments/$LINEA2" ] && rm -f "$MEETING_DIR/Attachments/$LINEA2"
|
||||
done < ./tmp/attachment_urls 3< ./tmp/attachment_names
|
||||
echo "All attachments saved."
|
||||
|
||||
@ -399,6 +433,11 @@ while read -r INDEX_URL_PRE CITY_ARCHIVE_NAME_PRE CALENDAR_NAME_PRE; do
|
||||
fi
|
||||
|
||||
echo "All files from this meeting have been saved."
|
||||
find "$MEETING_DIR" -type f -size 0 -delete
|
||||
echo "Cleaning PDFs for archive.org..."
|
||||
find "$MEETING_DIR" -type f -name '*.pdf' -print0 | xargs -0 -n1 qpdf --replace-input
|
||||
# qpdf repairs and leaves garbage original PDFs
|
||||
find "$MEETING_DIR" -type f -name '*~qpdf-orig' -delete -print
|
||||
fi
|
||||
|
||||
((i++))
|
||||
@ -410,14 +449,12 @@ while read -r INDEX_URL_PRE CITY_ARCHIVE_NAME_PRE CALENDAR_NAME_PRE; do
|
||||
GREPLIST=$(echo $LINE | grep 'class="MeetingTypeListbox"')
|
||||
if [[ "$GREPLIST" != "" ]]; then
|
||||
echo "SCRAPE_ESCRIBE: Found meeting type list."
|
||||
FOUNDLIST=1
|
||||
FOUNDLIST="TRUE"
|
||||
fi
|
||||
done < $INDEX_PAGE
|
||||
else
|
||||
INDEX_END=1
|
||||
INDEX_END="TRUE"
|
||||
echo "SCRAPE_ESCRIBE: Couldn't save index!"
|
||||
fi
|
||||
done
|
||||
done < websites.csv
|
||||
|
||||
|
||||
|
||||
Loading…
Reference in New Issue
Block a user