google slides works enough for search

.gitignore

@@ -1,9 +1,9 @@
 **/*.sw*
-spike/review/reinvent/ezmded/server/ezmded
-spike/review/reinvent/ezmded/server/server
-spike/review/reinvent/ezmded/server/testdata/files/**/*
-spike/review/reinvent/ezmded/server/testdata/workd/**/*
-spike/review/reinvent/ezmded/server/testdata/media/**/*
-spike/review/reinvent/ezmded/server/testdata/index.html
-spike/review/reinvent/ezmded/ui/render
-spike/review/reinvent/ezmded/ui/**/.*.html
+server/ezmded
+server/server
+server/testdata/files/**/*
+server/testdata/workd/**/*
+server/testdata/media/**/*
+server/testdata/index.html
+ui/render
+ui/**/.*.html

app/ui/run.sh

@@ -1 +0,0 @@
-../../spike/review/run.sh

crawler/google.sh

@@ -0,0 +1,76 @@
+#! /bin/bash
+
+google() (
+	_is_slides() {
+		echo "$@" | grep -q 'docs.google.com.presentation'
+	}
+
+	_is_sheets() {
+		echo "$@" | grep -q 'docs.google.com.spreadsheets'
+	}
+
+	_is_doc() {
+		echo "$@" | grep -q 'docs.google.com.document'
+	}
+
+   is() {
+		_is_sheets "$@" || _is_doc "$@" || _is_slides "$@"
+   }
+
+	human_url() {
+		echo "$1"
+	}
+
+   get() {
+		local url="$1"
+		local id="${url%/*}"
+		id="${id##*/}"
+		local downloaded="$(rclone get_google "$id")"
+		echo "# ${downloaded##*/}"
+		echo ""
+		if [ "${downloaded##*.}" == "csv" ]; then
+			_csv_to_md "$downloaded"
+		elif [ "${downloaded##*.}" == "html" ]; then
+			_html_to_md "$downloaded"
+		else
+			cat "$downloaded"
+		fi
+   }
+
+	_html_to_md() {
+		which pandoc &> /dev/null
+		local f="$1"
+		#log f=$f
+		cat "$f" \
+			| sed 's/.*<body/<body/' \
+			| sed 's/<\/body>.*/<\/body>/' \
+			| sed 's/<[\/]*span[^>]*>//g' \
+			| perl -pe 's|<div class="c[0-9][0-9]*">.*?<\/div>||g' \
+			| sed 's/<\([a-z][a-z]*\)[^>]*/<\1/g' \
+			| pandoc - -f html -t commonmark -s -o - \
+			| sed 's/^<[\/]*div>$//g'
+	}
+
+	_csv_to_md() {
+		local f="$1"
+		(
+			head -n 1 "$f"
+			head -n 1 "$f" \
+				| sed 's/^[^,][^,]*/--- /' \
+				| sed 's/[^,][^,]*$/ ---/' \
+				| sed 's/,[^,][^,]*/, --- /g' \
+				| sed 's/[^|]$/|/'
+			tail -n +2 "$f"
+		) \
+			| grep . \
+			| sed 's/,/ | /g' \
+			| sed 's/^/| /'
+	}
+
+   expand() {
+		get "$@" | head -n 1 | sed 's/^[#]* //' | base64
+   }
+
+   "$@"
+)
+

crawler/main.sh

@@ -4,6 +4,7 @@ main() {
 	config
 	log crawling ids...
 	for id in $(crawlable_ids); do
+		log crawling id $id
 		crawl "$id"
 	done
 	log rewriting ids...
@@ -20,8 +21,12 @@ config() {
 	export CACHE_DURATION=$((60*50))
 	export NOTES_ADDR="${NOTES_ADDR:-"http://localhost:3004"}"
 	export GITLAB_PAT="$GITLAB_PAT"
+	export RCLONE_CONFIG="$RCLONE_CONFIG"
+	export RCLONE_CONFIG_PASS="$RCLONE_CONFIG_PASS"
 	source ./gitlab.sh
 	source ./gitlab_wiki.sh
+	source ./google.sh
+	source ./rclone.sh
 	source ./cache.sh
 	source ./notes.sh
 }
@@ -56,12 +61,7 @@ crawlable_ids() {
 }
 
 crawl() {
-	local cache_key="crawled $*"
-	# TODO
-	if false && cache get "$cache_key"; then
-		return
-	fi
-	_crawl "$@" | cache put "$cache_key"
+	_crawl "$@"
 }
 
 _crawl() {
@@ -73,7 +73,7 @@ _crawl() {
 			"$id"
 	)"
 	local crawlable_source="$(extract_crawlable_source "$content")"
-	for backend in gitlab gitlab_wiki; do
+	for backend in gitlab gitlab_wiki google; do
 		if $backend is "$crawlable_source"; then
 			crawl_with $backend "$json"
 			return $?
@@ -149,6 +149,7 @@ crawl_with() {
 		ID="${ID%/}"
 		log "	$ID ($TITLE): ${#CONTENT}"
 		push_crawled "$ID" "$TITLE" "$CONTENT"
+		log "	/$ID ($TITLE): ${#CONTENT}"
 	}
 	if [ "${#expanded[@]}" -gt 0 ]; then
 		for i in $(seq 0 $(("${#expanded[@]}"-1))); do
@@ -166,7 +167,7 @@ push_crawled() {
 is_crawlable() {
 	local crawlable_source="$(extract_crawlable_source "$*")"
 	# https://unix.stackexchange.com/questions/181254/how-to-use-grep-and-cut-in-script-to-obtain-website-urls-from-an-html-file
-	local url_pattern="(http|https)://[a-zA-Z0-9./?=_%:-]*"
+	local url_pattern="(http|https)://[a-zA-Z0-9./?=_%:\-\#--]*"
 	echo "$crawlable_source" | cut -c 1-300 | grep -q -E "^[ ]*$url_pattern[ ]*$"
 }
 

crawler/notes.sh

@@ -2,11 +2,15 @@
 
 notes() (
 	ids() {
-		_recurse_ids "" "$(_tree)"
+		_recurse_ids "$(_tree)"
 	}
 
 	_tree() {
-      __tree "$@"
+		local cache_key="notes _tree"
+		if CACHE_DURATION=5 cache get "$cache_key"; then
+			return 0
+		fi
+      __tree "$@" | cache put "$cache_key"
 	}
 
 	__tree() {
@@ -18,8 +22,7 @@ notes() (
 	}
 
 	_recurse_ids() {
-		local prefix="$1"
-		local json="$2"
+		local json="$1"
 		if echo "$json" | jq .Branches | grep -q ^null$; then
 			return 0
 		fi
@@ -29,22 +32,32 @@ notes() (
 		fi
 		for line in $b64lines; do
 			line="$(echo "$line" | base64 --decode)"
-			local subfix="$(printf "%s/%s" "$prefix" "$line")"
-			subfix="${subfix#/}"
-			if ! _is_deleted "$subfix"; then
-				echo "$subfix"
+			if ! _is_deleted "$line"; then
+				echo "$line"
+				_recurse_ids "$(echo "$json" | jq -c ".Branches[\"$line\"]")"
 			fi
-			_recurse_ids "$subfix" "$(echo "$json" | jq -c ".Branches[\"$line\"]")"
 		done
 	}
 
    meta() {
+		local cache_key="notes meta $*"
+		if CACHE_DURATION=5 cache get "$cache_key"; then
+			return 0
+		fi
+		_meta "$@" | cache put "$cache_key"
+	}
+
+   _meta() {
 		local id="$1"
 		local tree="$(_tree)"
-		for subid in ${id//\// }; do
-			tree="$(echo "$tree" | jq -c .Branches | jq -c ".[\"$subid\"]")"
+		local pid="${id%%/*}"
+		while [ "$id" != "$pid" ]; do
+			tree="$(echo "$tree" | jq ".Branches[\"$pid\"]")"
+			local to_add="${id#$pid/}"
+			to_add="${to_add%%/*}"
+			pid="$pid/$to_add"
 		done
-		echo "$tree" | jq .Leaf
+		echo "$tree" | jq ".Branches[\"$id\"].Leaf"
    }
 
    _is_deleted() {
@@ -90,11 +103,11 @@ notes() (
 		local id="$1"
 		local title="$2"
 		local body="$3"
-      echo "$body" | _nncurl \
+      _nncurl \
          -X PUT \
 			-H "Title: $title" \
          -d "$body" \
-         $NOTES_ADDR/api/v0/files/$id
+         $NOTES_ADDR/api/v0/files/$id >&2
    }
 
    "$@"

crawler/rclone.sh

@@ -0,0 +1,62 @@
+#! /bin/bash
+
+rclone() (
+	get_google() {
+      local cache_key="rclone get google 2 $*"
+      if cache get "$cache_key"; then
+         return 0
+      fi
+      _get_google "$@" | cache put "$cache_key"
+	}
+
+	_get_google() {
+		_rate_limit
+		local id="$1"
+		local out="$(mktemp -d)"
+		_cmd backend copyid work-notes-google: --drive-export-formats=csv,html,txt "$id" "$out/"
+		find "$out" -type f
+	}
+
+	_rate_limit() {
+		local f="/tmp/rclone.rate.limit"
+		local last=0
+		if [ -f "$f" ]; then
+			last="$(date -r "$f" +%s)"
+		fi
+		local now="$(date +%s)"
+		local since_last=$((now-last))
+		if ((since_last>2)); then
+			dur=-2
+		fi
+		dur=$((dur+2))
+		sleep $dur
+		touch "$f"
+	}
+
+	_ensure() {
+		which rclone &> /dev/null && rclone version &> /dev/null
+	}
+
+	_cmd() {
+		_ensure_google_config
+		__cmd "$@"
+   }
+
+	__cmd() {
+		_ensure
+   	RCLONE_CONFIG_PASS="$RCLONE_CONFIG_PASS" \
+			$(which rclone) \
+   	      --config "$RCLONE_CONFIG" \
+   	      --size-only \
+   	      --fast-list \
+   	      --retries 10 \
+   	      --retries-sleep 10s \
+   	   "$@"
+	}
+
+	_ensure_google_config() {
+		__cmd config show | grep -q work-notes-google
+	}
+
+	"$@"
+)

server/go.mod

@@ -0,0 +1,31 @@
+module ezmded
+
+go 1.17
+
+require (
+	github.com/google/uuid v1.3.0
+	go.mongodb.org/mongo-driver v1.7.2
+	gopkg.in/yaml.v2 v2.4.0
+	local/args v0.0.0-00010101000000-000000000000
+	local/gziphttp v0.0.0-00010101000000-000000000000
+	local/router v0.0.0-00010101000000-000000000000
+	local/simpleserve v0.0.0-00010101000000-000000000000
+)
+
+require github.com/go-stack/stack v1.8.0 // indirect
+
+replace local/args => ../../../../args
+
+replace local/logb => ../../../../logb
+
+replace local/storage => ../../../../storage
+
+replace local/router => ../../../../router
+
+replace local/simpleserve => ../../../../simpleserve
+
+replace local/gziphttp => ../../../../gziphttp
+
+replace local/notes-server => ../../../../notes-server
+
+replace local/oauth2 => ../../../../oauth2

server/todo.yaml

@@ -1,15 +1,15 @@
 todo:
+- scrape odo
+- mark generated via meta so other files in the dir can be created, deleted, replaced safely
+- rewrite links if available to local
 - table of contents
 - anchor per line
-- scrape odo
-- scrape gdoc
-- scrape gsheet
-- scrape gslide
 - anchor links work
-- rewrite links if available to local
 - ui; last updated; 2022.02.01T12:34:56
-- mark generated via meta so other files in the dir can be created, deleted, replaced safely
 done:
+- scrape gslide
+- scrape gsheet
+- scrape gdoc
 - alert box; https://concisecss.com/documentation/ui
 - hide checkbox for tree
 - do not rewrite .md title vs. link cause hrefs to ./gobs.md wont work

server/tree.go

@@ -1,6 +1,7 @@
 package main
 
 import (
+	"encoding/json"
 	"io/ioutil"
 	"os"
 	"path"
@@ -53,7 +54,6 @@ func (base Leaf) Merge(updated Leaf) Leaf {
 
 type Tree struct {
 	root string
-	cachedRoot Branch
 }
 
 func NewTree(root string) Tree {
@@ -62,25 +62,72 @@ func NewTree(root string) Tree {
 
 func (tree Tree) WithRoot(root string) Tree {
 	tree.root = root
-	tree.cachedRoot = Branch{}
 	return tree
 }
 
 func (tree Tree) GetRootMeta() (Branch, error) {
-	return tree.getRoot(NewID(""), false, false)
+	if meta, ok := tree.getCachedRootMeta(); ok {
+		return meta, nil
+	}
+	got, err := tree.getRoot(NewID(""), false, false)
+	if err != nil {
+		return Branch{}, err
+	}
+	tree.cacheRootMeta(got)
+	return got, err
 }
 
 func (tree Tree) GetRoot() (Branch, error) {
-	if !tree.cachedRoot.IsZero() {
-		return tree.cachedRoot, nil
+	if root, ok := tree.getCachedRoot(); ok {
+		return root, nil
 	}
 	got, err := tree.getRoot(NewID(""), true, false)
-	if err == nil {
-		tree.cachedRoot = got
+	if err != nil {
+		return Branch{}, err
 	}
+	tree.cacheRoot(got)
 	return got, err
 }
 
+func (tree Tree) getCachedRoot() (Branch, bool) {
+	return tree.getCachedFrom("root.json")
+}
+
+func (tree Tree) getCachedRootMeta() (Branch, bool) {
+	return tree.getCachedFrom("root_meta.json")
+}
+
+func (tree Tree) getCachedFrom(name string) (Branch, bool) {
+	b, err := ioutil.ReadFile(path.Join(tree.root, name))
+	if err != nil {
+		return Branch{}, false
+	}
+	var branch Branch
+	err = json.Unmarshal(b, &branch)
+	return branch, err == nil
+}
+
+func (tree Tree) cacheRoot(branch Branch) {
+	tree.cacheRootFrom("root.json", branch)
+}
+
+func (tree Tree) cacheRootMeta(branch Branch) {
+	tree.cacheRootFrom("root_meta.json", branch)
+}
+
+func (tree Tree) cacheRootFrom(name string, branch Branch) {
+	b, err := json.Marshal(branch)
+	if err != nil {
+		return
+	}
+	ensureAndWrite(path.Join(tree.root, name), b)
+}
+
+func (tree Tree) cacheClear() {
+	os.Remove(path.Join(path.Join(tree.root, "root.json")))
+	os.Remove(path.Join(path.Join(tree.root, "root_meta.json")))
+}
+
 func (tree Tree) getRoot(pid ID, withContent, withDeleted bool) (Branch, error) {
 	m := Branch{Branches: map[ID]Branch{}}
 	entries, err := os.ReadDir(tree.root)
@@ -92,7 +139,7 @@ func (tree Tree) getRoot(pid ID, withContent, withDeleted bool) (Branch, error)
 	}
 	for _, entry := range entries {
 		if entry.Name() == "data.yaml" {
-			if b, err := ioutil.ReadFile(path.Join(tree.root, entry.Name())); err != nil {
+			if b, err := peekLeaf(withContent, path.Join(tree.root, entry.Name())); err != nil {
 				return Branch{}, err
 			} else if err := yaml.Unmarshal(b, &m.Leaf); err != nil {
 				return Branch{}, err
@@ -115,6 +162,10 @@ func (tree Tree) getRoot(pid ID, withContent, withDeleted bool) (Branch, error)
 	return m, nil
 }
 
+func peekLeaf(all bool, path string) ([]byte, error) {
+	return ioutil.ReadFile(path)
+}
+
 func (tree Tree) toDir(id ID) string {
 	return path.Dir(tree.toData(id))
 }
@@ -124,6 +175,7 @@ func (tree Tree) toData(id ID) string {
 }
 
 func (tree Tree) Put(id ID, input Leaf) error {
+	tree.cacheClear()
 	if _, err := os.Stat(tree.toData(id)); os.IsNotExist(err) {
 		b, err := yaml.Marshal(Leaf{})
 		if err != nil {
@@ -144,11 +196,11 @@ func (tree Tree) Put(id ID, input Leaf) error {
 	if err := ensureAndWrite(tree.toData(id), b); err != nil {
 		return err
 	}
-	tree.cachedRoot = Branch{}
 	return nil
 }
 
 func (tree Tree) Del(id ID) error {
+	tree.cacheClear()
 	got, err := tree.Get(id)
 	if os.IsNotExist(err) {
 		return nil
@@ -164,8 +216,8 @@ func (tree Tree) Del(id ID) error {
 }
 
 func (tree Tree) HardDel(id ID) error {
+	tree.cacheClear()
 	os.RemoveAll(tree.toDir(id))
-	tree.cachedRoot = Branch{}
 	return nil
 }
 

spike/review/reinvent/ezmded/server/go.mod

@@ -1,31 +0,0 @@
-module ezmded
-
-go 1.17
-
-require (
-	github.com/google/uuid v1.3.0
-	go.mongodb.org/mongo-driver v1.7.2
-	gopkg.in/yaml.v2 v2.4.0
-	local/args v0.0.0-00010101000000-000000000000
-	local/gziphttp v0.0.0-00010101000000-000000000000
-	local/router v0.0.0-00010101000000-000000000000
-	local/simpleserve v0.0.0-00010101000000-000000000000
-)
-
-require github.com/go-stack/stack v1.8.0 // indirect
-
-replace local/args => ../../../../../../../../args
-
-replace local/logb => ../../../../../../../../logb
-
-replace local/storage => ../../../../../../../../storage
-
-replace local/router => ../../../../../../../../router
-
-replace local/simpleserve => ../../../../../../../../simpleserve
-
-replace local/gziphttp => ../../../../../../../../gziphttp
-
-replace local/notes-server => ../../../../../../../../notes-server
-
-replace local/oauth2 => ../../../../../../../../oauth2

ui/templates/_editor.ctmpl

@@ -113,7 +113,7 @@
 		element: document.getElementById('my-text-area'),
 		forceSync: true,
 		indentWithTabs: false,
-		initialValue: "loading...",
+		initialValue: "{{ .This.Content }}",
 		showIcons: ["code", "table"],
 		spellChecker: false,
 		sideBySideFullscreen: false,
@@ -133,6 +133,5 @@
 		},
 		status: ["lines", "words", "cursor"],
 	})
-	easyMDE.value({{ .This.Content }})
 </script>
 {{ end }}

ui/templates/_import.ctmpl

@@ -75,6 +75,9 @@
 	.tb_fullscreen {
 		height: 100%;
 	}
+	.button, button, input[type="button"] {
+		height: auto;
+	}
 </style>
 <script>
 	function http(method, remote, callback, body, headers) {