vacancies: Links in umega and elc getters

1 files changed, 83 insertions, 0 deletions

diff --git a/modules/vacancies/getter-helpers/umega-like.sh b/modules/vacancies/getter-helpers/umega-like.sh
new file mode 100755
index 0000000..061476b
--- /dev/null
+++ b/modules/vacancies/getter-helpers/umega-like.sh
@@ -0,0 +1,83 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+URL=$1
+
+mapfile inputlines <<<"$(elinks "$1")"
+
+
+## Collect links
+
+declare -A links
+inlinkssection=0
+for ((i=0; i<${#inputlines[@]}; i++)); do
+  line=$(tr -d $'\n' <<<"${inputlines[$i]}")
+  if grep '^\s*Visible links$' <<<"$line" >/dev/null; then
+    inlinkssection=1
+  elif [[ $inlinkssection -eq 1 ]] && grep '^\s*[0-9][0-9]*\. ' <<<"$line" >/dev/null; then
+    linknumber=$(sed 's/^\s*\([0-9]*\)\..*/\1/' <<<"$line")
+    linktext=$(sed 's/[^.]*\. //' <<<"$line")
+    links[$linknumber]=$linktext
+  fi
+done
+
+
+## Collect ads
+
+totalnum=
+counted=0
+
+accum=
+
+# 0: outside
+# 1: in overview images section
+# 2: in body
+state=0
+
+adlink=  # link for the ad currently being collected
+
+function writeout() {
+  if [[ $state -eq 2 && -n $accum ]]; then
+    if [[ -n $adlink ]]; then echo "$accum ($adlink)"; else echo "$accum"; fi
+    let counted+=1
+  fi
+  accum=
+}
+
+for ((i=0; i<${#inputlines[@]}; i++)); do
+  line=$(tr -d $'\n' <<<"${inputlines[$i]}")
+
+  # echo >&2 "?$state <$line>"
+  m=$(sed -n 's/[^0-9]*\([0-9]*\) *[Pp]roperties for rent.*/\1/p' <<<"$line")
+  if [[ -n $m ]]; then
+    totalnum=$m
+    continue
+  fi
+
+  if grep 'Overview image' <<<"$line" >/dev/null; then
+    # echo >&2 "overview <$line>"
+    writeout
+    state=1
+  elif [[ -z "$(tr -d '[:space:]' <<<"$line")" ]]; then
+    # echo >&2 "empty <$line>"
+    writeout
+    state=0
+    if [[ $counted -gt 0 ]]; then break; fi
+  elif [[ $state -eq 1 ]] && grep '^\s*\[' <<<"$line" >/dev/null; then
+    # echo >&2 "bodystart <$line>"
+    state=2
+    linknumber=$(sed 's/^\s*\[\([0-9]*\).*/\1/' <<<"$line")
+    if [[ -n ${links[$linknumber]:+1} ]]; then adlink=${links[$linknumber]}; fi
+    accum=$(sed 's/^\s*\[[0-9]*\]//' <<<"$line")
+  elif [[ $state -eq 2 ]]; then
+    # echo >&2 "bodycont <$line>"
+    # c2 a0 is UTF-8 for a non-breaking space
+    accum="$accum; $(sed 's/^\s*//; s/^\xc2\xa0//; s/\s*$//' <<<"$line")"
+  fi
+done
+
+writeout
+
+if [[ $counted != "$totalnum" ]]; then
+  echo "SCRIPT BROKEN, OFFERS MAY BE MISSED"
+fi