vacancies: elc-rentHEADmaster

1 files changed, 52 insertions, 0 deletions

diff --git a/modules/vacancies/getters/elc-rent.sh b/modules/vacancies/getters/elc-rent.sh
new file mode 100755
index 0000000..624e99e
--- /dev/null
+++ b/modules/vacancies/getters/elc-rent.sh
@@ -0,0 +1,52 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+totalnum=
+counted=0
+
+accum=
+
+# 0: outside
+# 1: in overview images section
+# 2: in body
+state=0
+
+mapfile inputlines <<<"$(elinks 'https://properties.edinburghlettingcentre.com/?per_page=60&address=Edinburgh&price_min&price_max=1200&bedrooms_min=1')"
+
+for ((i=0; i<${#inputlines[@]}; i++)); do
+  line=$(tr -d $'\n' <<<"${inputlines[$i]}")
+
+  # echo >&2 "?$state <$line>"
+  m=$(sed -n 's/[^0-9]*\([0-9]*\) *Properties 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>"
+    if [[ $state -eq 2 && -n $accum ]]; then echo "$accum"; let counted+=1; fi
+    accum=
+    state=1
+  elif [[ -z "$(tr -d '[[:space:]]' <<<"$line")" ]]; then
+    # echo >&2 "empty <$line>"
+    if [[ $state -eq 2 && -n $accum ]]; then echo "$accum"; let counted+=1; fi
+    accum=
+    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
+    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
+
+if [[ -n $accum ]]; then echo "$accum"; let counted+=1; fi
+
+if [[ $counted != $totalnum ]]; then
+  echo "SCRIPT BROKEN, OFFERS MAY BE MISSED"
+fi