Merge from develop

docs/TODO.md

@@ -6,10 +6,10 @@
   * should only crawl these pages once
   * but should be included in the links part
 * when sorting links, "uniq" them as well
+* be smart about index.html?
 
 ## Cleanup
 
-* rename Page -> crawlerState
 * helper function to do the whole crawl process
 * document code better
   * test with godoc
@@ -17,7 +17,6 @@
 
 ## Features
 
-* some sort of status during crawl?
 * command line options, flag parsing
 * check RFCs on URL character sets, compliance, etc.
 * packagize parts of app

src/docrawler/docrawler.go

@@ -2,84 +2,120 @@ package main
 
 import (
 	"fmt"
+	"log"
 	"os"
+	"time"
 )
 
-// docrawl begins crawling the site at "url"
+// docrawl begins crawling the site at "homeurl"
-func docrawl(url string) []*Page {
+func docrawl(homeurl string) itemSlice {
-	// our result structure
+	// set of what we have already crawled
-	var sitemap []*Page
+	crawled := make(map[string]struct{})
 
-	// what we've already parsed, what we still need to parse
+	// map (of urlString -> httpItem) of our results
-	crawled := make(map[string]*Page)
+	results := make(itemMap)
-	queued := make(map[string]*Page)
 
-	// create first page and add to the queue
+	// a channel to receive our results on
-	homepage, _ := NewPage(nil, url) // TODO check error
+	rchan := make(chan *httpItem)
-	queued[homepage.Identifier] = homepage
 
-	// main loop
+	// start a ticker which we'll use to output status and check for completion
-	for len(queued) > 0 {
+	ticker := time.Tick(250 * time.Millisecond)
-		// pop a page off of queued
+
-		var dest *Page
+	// create first page's httpItem
-		var destURL string
+	homeitem, err := newHTTPItem(nil, homeurl)
-		for destURL, dest = range queued {
+	if err != nil {
-			break
+		log.Fatal("unable to create httpItem for homeurl")
 	}
-		delete(queued, destURL)
 
-		// mark this page as complete
+	// set our number of outstanding pages (initially 1 to account for first page)
-		crawled[dest.Identifier] = dest
+	crawlingCount := 1
 
-		// add page to list
+	// start the home page crawl
-		sitemap = append(sitemap, dest)
+	crawled[homeitem.url.String()] = struct{}{}
+	go crawlItem(homeitem, rchan)
 
-		// make sure this URL has the same hostname as our first page
+	// wait for results
-		if dest.URL.Host != homepage.URL.Host {
+	for {
-			// skip URLs associated with other Hosts
+		select {
-			dest.Skipped = true
+		case r := <-rchan: // new results?
+			// add result to our results map
+			results[r.url.String()] = r
+
+			// decrease the outstanding page count by 1
+			crawlingCount--
+
+			// start crawly any new child pages we haven't yet crawled
+			for i, c := range r.children {
+				// see if we already have a result for this page, if so, point to that result
+				if existing, ok := results[c.url.String()]; ok {
+					r.children[i] = existing
 					continue
 				}
 
+				// see if we're already crawling this page (but maybe don't have results yet)
+				if _, ok := crawled[c.url.String()]; !ok {
+					// haven't crawled this one yet, do so now
+					crawlingCount++
+					crawled[c.url.String()] = struct{}{}
+					go crawlItem(c, rchan)
+				}
+			}
+
+		case <-ticker: // our regular ticker. for status output and checking for completion.
+			// output status to console
+			log.Printf("Crawled %v links, have %v left.\n", len(results), crawlingCount)
+
+			// see if we're finished
+			if crawlingCount == 0 {
+				// finished! convert results map to a slice and return it
+				rslice := itemSlice{}
+				for _, v := range results {
+					rslice = append(rslice, v)
+				}
+				return rslice
+			}
+		}
+	}
+}
+
+// crawlItem crawls a single httpItem, fetching the header, hte page, parsing it,
+// and filling out its structure as much as possible
+func crawlItem(item *httpItem, rchan chan<- *httpItem) {
+	// make sure this item is the same domain (i.e. URL "host part") as its referrer
+	if item.refurl != nil && item.url.Host != item.refurl.Host {
+		// skip URLs associated with other Hosts
+		item.linkType = tRemote
+		rchan <- item
+		return
+	}
+
 	// fetch page
-		text, err := fetchPage(dest)
+	text, err := fetchPage(item)
 	if err != nil {
-			dest.Broken = true
+		rchan <- item
-			continue
+		return
 	}
 
 	// parse links
-		title, links, err := parseLinks(text)
+	title, links, _ := parseLinks(text) // TODO remove error from parseLinks, not needed
 	if err != nil {
-			dest.Broken = true
+		item.linkType = tBroken
-			continue
+		rchan <- item
+		return
 	}
-		dest.Title = title
+	item.title = title
+
+	// walk links and add them as children to the current item
 	for _, l := range links {
-			p, err := NewPage(dest, l)
+		newItem, err := newHTTPItem(item, l)
 		if err != nil {
-				continue // TODO check error
+			continue // TODO bad item
+		}
+		item.children = append(item.children, newItem)
 	}
 
-			// see if we already know about this page
+	// send back our item struct now that it's all filled out
-			pCrawled, haveCrawled := crawled[p.Identifier]
+	rchan <- item
-			pQueued, haveQueued := queued[p.Identifier]
-			if haveCrawled {
-				// add the previously crawled page to children
-				dest.Children = append(dest.Children, pCrawled)
-			} else if haveQueued {
-				// add the previously queued page to children
-				dest.Children = append(dest.Children, pQueued)
-			} else {
-				// add this new page (never seen) to children
-				dest.Children = append(dest.Children, p)
-				// .. and queue it up for crawling
-				queued[p.Identifier] = p
-			}
-		}
-	}
-
-	return sitemap
 }
 
 func main() {

src/docrawler/docrawler_test.go

@@ -58,9 +58,9 @@ func TestBadRequest(t *testing.T) {
 
 // TestSimpleFetch makes sure we can fetch a file and we get exactly what we expect
 func TestSimpleFetch(t *testing.T) {
-	page, err := NewPage(nil, baseURL+"fetch_test.html")
+	page, err := newHTTPItem(nil, baseURL+"fetch_test.html")
 	if err != nil {
-		t.Error("problem creating New Page struct")
+		t.Error("problem creating New httpItem struct")
 	}
 	body, err := fetchPage(page)
 	if err != nil {
@@ -82,19 +82,19 @@ func TestSimpleMap(t *testing.T) {
 	// cycle through every page, counting stuff, finding specific pages
 	// so that we can then verify certain aspects of each page
 	skipped, children := 0, 0
-	var homepage, aboutpage, skippage *Page
+	var homepage, aboutpage, skippage *httpItem
 	for _, p := range pages {
-		if p.Title == "Home" {
+		if p.title == "Home" {
 			homepage = p
-		} else if p.Title == "About Test" {
+		} else if p.title == "About Test" {
 			aboutpage = p
-		} else if p.URL.String() == "http://doesntexist23492387492837492374982734.com/" {
+		} else if p.url.String() == "http://doesntexist23492387492837492374982734.com/" {
 			skippage = p
 		}
-		if p.Skipped {
+		if p.linkType == tRemote {
 			skipped++
 		}
-		children += len(p.Children)
+		children += len(p.children)
 	}
 
 	// check page counts
@@ -112,44 +112,44 @@ func TestSimpleMap(t *testing.T) {
 	}
 
 	// check first page URL
-	if homepage.URL.String() != baseURL {
+	if homepage.url.String() != baseURL {
 		t.Error("page URL is invalid")
 	}
 
 	// specific page number of children
-	if len(homepage.Children) != 4 {
+	if len(homepage.children) != 4 {
-		t.Logf("got %v, wanted %v\n", len(homepage.Children), 4)
+		t.Logf("got %v, wanted %v\n", len(homepage.children), 4)
 		t.Fatal("got wrong number of links")
 	}
-	if len(aboutpage.Children) != 4 {
+	if len(aboutpage.children) != 4 {
-		t.Logf("got %v, wanted %v\n", len(aboutpage.Children), 4)
+		t.Logf("got %v, wanted %v\n", len(aboutpage.children), 4)
 		t.Fatal("got wrong number of links")
 	}
 
 	// make sure remote pages were skipped
-	if homepage.Skipped != false {
+	if homepage.linkType == tRemote {
-		t.Logf("found a page that should NOT have been skipped %q", homepage.Children[0].URL.String())
+		t.Logf("found a page that should NOT have been skipped %q", homepage.children[0].url.String())
 		t.Error("child was skipped")
 	}
-	if skippage.Skipped != true {
+	if skippage.linkType != tRemote {
-		t.Logf("found a page that should have been skipped %q", pages[3].Children[2].URL.String())
+		t.Logf("found a page that should have been skipped %q", pages[3].children[2].url.String())
 		t.Error("child wasn't skipped")
 	}
 
 	// verify page content
 	// TODO find a better way to do this
-	//if homepage.Children[0].URL.String() != baseURL+"about.html" {
+	//if homepage.children[0].url.String() != baseURL+"about.html" {
-	//t.Logf("   Got: %q\n", homepage.Children[0].URL.String())
+	//t.Logf("   Got: %q\n", homepage.children[0].url.String())
 	//t.Logf("Wanted: %q\n", baseURL+"about.html")
 	//t.Error("link name is incorrect")
 	//}
-	//if homepage.Children[1].URL.String() != baseURL+"assets/image.png" {
+	//if homepage.children[1].url.String() != baseURL+"assets/image.png" {
-	//t.Logf("   Got: %q\n", homepage.Children[1].URL.String())
+	//t.Logf("   Got: %q\n", homepage.children[1].url.String())
 	//t.Logf("Wanted: %q\n", baseURL+"assets/image.png")
 	//t.Error("link name is incorrect")
 	//}
-	//if homepage.Children[3].URL.String() != baseURL+"scripts/blah.js" {
+	//if homepage.children[3].url.String() != baseURL+"scripts/blah.js" {
-	//t.Logf("   Got: %q\n", homepage.Children[3].URL.String())
+	//t.Logf("   Got: %q\n", homepage.children[3].url.String())
 	//t.Logf("Wanted: %q\n", baseURL+"scripts/blah.js")
 	//t.Error("link name is incorrect")
 	//}
@@ -157,28 +157,28 @@ func TestSimpleMap(t *testing.T) {
 
 // TestHeaderFetching tests fetchFiletype() to see if we are getting expected results
 func TestHeaderFetching(t *testing.T) {
-	basepage, err := NewPage(nil, baseURL)
+	basepage, err := newHTTPItem(nil, baseURL)
 	if err != nil {
-		t.Error("problem creating New Page struct")
+		t.Error("problem creating New httpItem struct")
 	}
-	if err := fetchFiletype(basepage); err != nil || basepage.MediaType != "text/html" {
+	if err := fetchFiletype(basepage); err != nil || basepage.linkType != tHTMLPage {
 		t.Error("problem fetching filetype")
 	}
 
-	page, err := NewPage(basepage, "about.html")
+	page, err := newHTTPItem(basepage, "about.html")
 	if err != nil {
-		t.Error("problem creating New Page struct")
+		t.Error("problem creating New httpItem struct")
 	}
-	if err := fetchFiletype(page); err != nil || page.MediaType != "text/html" {
+	if err := fetchFiletype(page); err != nil || page.linkType != tHTMLPage {
 		t.Error("problem fetching filetype")
 	}
 
-	page, err = NewPage(basepage, "assets/image.png")
+	page, err = newHTTPItem(basepage, "assets/image.png")
 	if err != nil {
-		t.Error("problem creating New Page struct")
+		t.Error("problem creating New httpItem struct")
 	}
-	if err := fetchFiletype(page); err != nil || page.MediaType != "image/png" {
+	if err := fetchFiletype(page); err != nil || page.linkType != tAsset {
-		t.Logf("got %q, wanted %q", page.MediaType, "image/png")
+		t.Logf("got %q, wanted %q", page.linkType, tAsset)
 		t.Error("problem fetching filetype")
 	}
 }

src/docrawler/fetch.go

@@ -9,19 +9,22 @@ import (
 
 var (
 	errContentTypeNotFound = errors.New("no Content-Type header found")
-	errFetchError          = errors.New("couldn't fetch page")
+	errFetchError          = errors.New("couldn't fetch item")
+	errFileTypeUnknown     = errors.New("couldn't determine file type")
 )
 
 // fetchFiletype performs an http HEAD to get the media type, and sets it
-// directly in Page.MediaType
+// directly in httpItem.mediaType
-func fetchFiletype(page *Page) error {
+func fetchFiletype(item *httpItem) error {
-	resp, err := http.Head(page.URL.String())
+	resp, err := http.Head(item.url.String())
 	if err != nil {
+		item.linkType = tBroken
 		return err
 	}
 
 	// check response code
 	if resp.StatusCode != http.StatusOK {
+		item.linkType = tBroken
 		return errFetchError
 	}
 
@@ -29,35 +32,46 @@ func fetchFiletype(page *Page) error {
 	contentType, ok := resp.Header["Content-Type"]
 	if !ok {
 		// no Content-Type header found
+		item.linkType = tBroken
 		return errContentTypeNotFound
 	}
 
 	// parse mime type
 	mediatype, _, err := mime.ParseMediaType(contentType[0])
 	if err != nil {
+		item.linkType = tBroken
 		return err
 	}
 
-	// set and return
+	// success, set item type and return
-	page.MediaType = mediatype
+	if mediatype == "text/html" {
+		item.linkType = tHTMLPage
+	} else {
+		item.linkType = tAsset
+	}
 	return nil
 }
 
-// fetchPage takes a Page, GETs it, and returns the body as a string
+// fetchPage takes a httpItem, GETs it, and returns the body as a string
-func fetchPage(page *Page) (string, error) {
+func fetchPage(item *httpItem) (string, error) {
 	// figure out the file type
-	err := fetchFiletype(page)
+	err := fetchFiletype(item)
 	if err != nil {
 		return "", err
 	}
 
-	// we only want to fetch html
+	// make sure we got a file type!
-	if page.MediaType != "text/html" {
+	if item.linkType == tUnknown {
+		return "", errFileTypeUnknown
+	}
+
+	// we only want to fetch html, so return if it's anything else
+	if item.linkType != tHTMLPage {
 		return "", nil // but this isn't an error!
 	}
 
 	// GET the url
-	resp, err := http.Get(page.URL.String())
+	resp, err := http.Get(item.url.String())
 	if err != nil {
 		return "", err
 	}

src/docrawler/fetch_test.go

@@ -6,7 +6,7 @@ import (
 
 // TestFailedHeaderFetching tests fetchFiletype() to see if we are getting expected failures
 func TestFailedHeaderFetching(t *testing.T) {
-	page, err := NewPage(nil, "http://doesntexist23492387492837492374982734.com/")
+	page, err := newHTTPItem(nil, "http://doesntexist23492387492837492374982734.com/")
 	if err != nil {
 		t.Error("problem creating New Page struct")
 	}
@@ -17,7 +17,7 @@ func TestFailedHeaderFetching(t *testing.T) {
 
 // TestFailedFetching tests fetchFiletype() to see if we are getting expected failures
 func TestFailedFetching(t *testing.T) {
-	page, err := NewPage(nil, "http://doesntexist23492387492837492374982734.com/")
+	page, err := newHTTPItem(nil, "http://doesntexist23492387492837492374982734.com/")
 	if err != nil {
 		t.Error("problem creating New Page struct")
 	}

src/docrawler/item.go

@@ -0,0 +1,55 @@
+package main
+
+import (
+	"net/url"
+)
+
+// itemSlice is a convenience type for a slice of items
+type itemSlice []*httpItem
+
+// itemMap convenience type for map of string's (URL) to items
+type itemMap map[string]*httpItem
+
+// itemType is an enum so we know how to classify each item
+type itemType int
+
+const (
+	tUnknown itemType = iota
+	tHTMLPage
+	tAsset
+	tRemote
+	tBroken
+)
+
+// httpItem is a struct which defines a single page, which URLs (links and assets) it contains, etc.
+type httpItem struct {
+	url      *url.URL
+	refurl   *url.URL
+	title    string
+	linkType itemType
+	children itemSlice
+}
+
+// newHTTPItem takes a referring httpItem + a URL and returns a new &httpItem{}
+func newHTTPItem(referrer *httpItem, rawurl string) (*httpItem, error) {
+	// determine the base URL so we can resulve this into a full URL
+	baseURL := ""
+	if referrer != nil {
+		baseURL = referrer.url.String()
+	}
+
+	// resolve URL into a full absolute URL (non-relative)
+	u, err := resolveURL(baseURL, rawurl)
+	if err != nil {
+		return nil, err
+	}
+
+	// figure out referrer url
+	var rurl *url.URL
+	if referrer != nil {
+		rurl = referrer.url
+	}
+
+	// create struct and return
+	return &httpItem{url: u, refurl: rurl}, nil
+}

src/docrawler/item_test.go

@@ -0,0 +1,13 @@
+package main
+
+import (
+	"testing"
+)
+
+// TestBadnewHTTPItem tests newHTTPItem to see if we are getting expected failures
+func TestBadnewHTTPItem(t *testing.T) {
+	_, err := newHTTPItem(nil, "h%20t\tp://doesntexist23492387492837492374982734.com/")
+	if err == nil {
+		t.Error("tried creating a newHTTPItem with an invalid URL but didn't get an error")
+	}
+}

src/docrawler/output.go

@@ -15,32 +15,41 @@ type Location struct {
 	Remote []string
 }
 
-func sitemapToLocations(pages []*Page) []*Location {
+// implement Location slice sorting (by URL)
+type byURL []*Location
+
+// implement Sort interface on byURL (which is a []*Location)
+func (l byURL) Len() int           { return len(l) }
+func (l byURL) Less(i, j int) bool { return l[i].URL < l[j].URL }
+func (l byURL) Swap(i, j int)      { l[i], l[j] = l[j], l[i] }
+
+// sitemapToLocations converts an itemSlice to []*Location, which is appropriate for marshalling
+func sitemapToLocations(pages itemSlice) []*Location {
 	// build a map of our pages
-	pageMap := make(map[string]*Page)
+	pageMap := make(map[string]*httpItem)
 	for _, p := range pages {
-		pageMap[p.URL.String()] = p
+		pageMap[p.url.String()] = p
 	}
 
 	// build a slice of locations (one per page)
 	var locations []*Location
 	for _, p := range pages {
-		if p.MediaType == "text/html" {
+		if p.linkType == tHTMLPage {
 			// create a location for this page
-			l := &Location{URL: p.URL.String(), Title: p.Title}
+			l := &Location{URL: p.url.String(), Title: p.title}
 
 			// add its children
-			for _, c := range p.Children {
+			for _, c := range p.children {
 				// look up this child's media type from the root list of pages
-				//mediaType := pageMap[c.URL.String()].MediaType
+				//mediaType := pageMap[c.url.String()].mediaType
-				if c.Skipped {
+				if c.linkType == tRemote {
-					l.Remote = append(l.Remote, c.URL.String())
+					l.Remote = append(l.Remote, c.url.String())
-				} else if c.MediaType == "text/html" {
+				} else if c.linkType == tHTMLPage {
-					l.Links = append(l.Links, c.URL.String())
+					l.Links = append(l.Links, c.url.String())
-				} else if c.Broken {
+				} else if c.linkType == tBroken {
-					l.Broken = append(l.Broken, c.URL.String())
+					l.Broken = append(l.Broken, c.url.String())
 				} else {
-					l.Assets = append(l.Assets, c.URL.String())
+					l.Assets = append(l.Assets, c.url.String())
 				}
 			}
 
@@ -59,6 +68,8 @@ func sitemapToLocations(pages []*Page) []*Location {
 		}
 	}
 
+	// sort the Locations themselves and return
+	sort.Sort(byURL(locations))
 	return locations
 }
 

src/docrawler/page_test.go

@@ -1,13 +0,0 @@
-package main
-
-import (
-	"testing"
-)
-
-// TestBadNewPage tests NewPage to see if we are getting expected failures
-func TestBadNewPage(t *testing.T) {
-	_, err := NewPage(nil, "h%20t\tp://doesntexist23492387492837492374982734.com/")
-	if err == nil {
-		t.Error("tried creating a NewPage with an invalid URL but didn't get an error")
-	}
-}