Flesh out README and documentation

LICENSE.txt

@@ -0,0 +1,14 @@
+            DO WHAT THE FUCK YOU WANT TO PUBLIC LICENSE
+                    Version 2, December 2004
+
+ Copyright (C) 2004 Sam Hocevar <sam@hocevar.net>
+
+ Everyone is permitted to copy and distribute verbatim or modified
+ copies of this license document, and changing it is allowed as long
+ as the name is changed.
+
+            DO WHAT THE FUCK YOU WANT TO PUBLIC LICENSE
+   TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
+
+  0. You just DO WHAT THE FUCK YOU WANT TO.
+

README.md

@@ -0,0 +1,44 @@
+# DeadLinks
+
+A tool for crawling and finding links to URLs which no longer exist. deadlinks
+supports the HTTP(s) and gemini protocols, and is intended for periodically
+checking links on personal websites and blogs.
+
+## Library
+
+The `deadlinks` package is designed to be easily embedded into a process and
+have its results displayed in something like a status page.
+
+[See the godocs for more info.](https://godocs.io/code.betamike.com/mediocregopher/deadlinks)
+
+## Command-Line
+
+The command-line utility can be installed using `go install`:
+
+```
+go install code.betamike.com/mediocregopher/deadlinks/cmd/deadlinks
+```
+
+The `-urls` parameter is required. Given one or more URLs it will check each one
+for any dead links:
+
+```
+deadlinks -urls 'https://mediocregopher.com,gemini://mediocregopher.com'
+```
+
+Any links which are dead will be output to stdout as YAML objects, each
+containing the dead URL, the error encountered, and which pages link to it.
+
+In order to recursively crawl through links you can give one or more regex
+patterns. Any URL which matches a pattern will have its links checked as well
+(and if any of those link URLs match a pattern their links will be checked, and
+so on):
+
+```
+deadlinks \
+    -urls 'https://mediocregopher.com,gemini://mediocregopher.com' \
+    -patterns '://mediocregopher.com'
+```
+
+There are further options available which affect the utility's behavior, see
+`deadlinks -h` for more.

client.go

@@ -78,6 +78,7 @@ type client struct {
 //
 // Some schemas automatically return success:
 // - mailto
+// - data
 func NewClient(opts *ClientOpts) Client {
 	return &client{*opts.withDefaults()}
 }
@@ -261,7 +262,7 @@ func (c *client) get(
 		return c.getGemini(ctx, url, redirectDepth)
 	case "http", "https":
 		return c.getHTTP(ctx, url, redirectDepth)
-	case "mailto":
+	case "mailto", "data":
 		return c.noOpGet()
 	default:
 		return "", nil, fmt.Errorf("unsupported scheme %q", scheme)

cmd/deadlinks/main.go

@@ -0,0 +1,97 @@
+package main
+
+import (
+	"context"
+	"flag"
+	"io"
+	"log"
+	"os"
+	"os/signal"
+	"runtime"
+	"strings"
+	"time"
+
+	"code.betamike.com/mediocregopher/deadlinks"
+	"code.betamike.com/mediocregopher/mediocre-go-lib/miter"
+	"gopkg.in/yaml.v3"
+)
+
+type loggingClient struct {
+	inner deadlinks.Client
+}
+
+func (c loggingClient) Get(
+	ctx context.Context, url deadlinks.URL,
+) (
+	string, io.ReadCloser, error,
+) {
+	log.Printf("querying %q", url)
+	return c.inner.Get(ctx, url)
+}
+
+func main() {
+	var (
+		storePath   = flag.String("store-path", "", "Path to sqlite storage file. If not given then a temporary in-memory storage is used")
+		maxAge      = flag.Duration("max-age", 0, "Maximum duration since last check of a resource, before it must be checked again. Must be used with -store-path")
+		urls        = flag.String("urls", "", `Comma-separated list of URLs which are always checked. At least one is required`)
+		patternsStr = flag.String("patterns", "", "Comma-separated list of regexps. All URLs which match one of these will have their links checked as well")
+		concurrency = flag.Int("concurrency", runtime.NumCPU()/2, "Number simultaneous requests to make at a time")
+	)
+
+	flag.Parse()
+
+	if *urls == "" {
+		log.Fatal("-urls is required")
+	}
+
+	var patterns []string
+	if *patternsStr != "" {
+		patterns = strings.Split(*patternsStr, ",")
+	}
+
+	ctx, cancel := signal.NotifyContext(context.Background(), os.Interrupt)
+	defer cancel()
+
+	store := deadlinks.NewSQLiteStore(&deadlinks.SQLiteStoreOpts{
+		Path: *storePath,
+	})
+	defer store.Close()
+
+	dl, err := deadlinks.New(
+		ctx,
+		store,
+		strings.Split(*urls, ","),
+		patterns,
+		&deadlinks.Opts{
+			NewClient: func() deadlinks.Client {
+				return loggingClient{deadlinks.NewClient(nil)}
+			},
+			Concurrency: *concurrency,
+			OnError: func(err error) {
+				log.Printf("runtime error: %v", err)
+			},
+		},
+	)
+
+	if err != nil {
+		log.Fatalf("initialization error: %v", err)
+	}
+
+	lastCheckedBefore := time.Now().Add(-*maxAge)
+
+	if err := dl.Update(ctx, lastCheckedBefore); err != nil {
+		log.Fatalf("update encountered error: %v", err)
+	}
+
+	enc := yaml.NewEncoder(os.Stdout)
+	defer os.Stdout.Sync()
+
+	iter := dl.GetByStatus(deadlinks.ResourceStatusError)
+	err = miter.ForEach(ctx, iter, func(r deadlinks.Resource) error {
+		return enc.Encode(r)
+	})
+
+	if err != nil {
+		log.Fatalf("iterating over errored resources failed: %v", err)
+	}
+}

deadlinks.go

@@ -1,6 +1,50 @@
 // Package deadlinks implements a liveness checker for hyperlinks in HTML and
 // gemtext documents.
 //
+// # URLs
+//
+// DeadLinks crawls and keeps track of hyperlinks between different
+// resources, such as webpages and gemtext documents. If a resource is not
+// linked to from any other resources then DeadLinks forgets about it.
+//
+// For this reason it is required to have a starting set of URLs which DeadLinks
+// will not forget about; these are the pinned URLs. Pinned URLs act as the
+// starting point for crawling.
+//
+// When DeadLinks traverses a URL link, it will check the liveness of that URL's
+// resource, but it will not by default recur into _that_ resource's links. It
+// will only do so if the URL matches one of the given regex patterns which
+// DeadLinks was configured with.
+//
+// # Basic Usage
+//
+// DeadLinks can be initialized using `New`:
+//
+//	store := deadlinks.NewSQLiteStore(nil)
+//	defer store.Close()
+//
+//	pinnedURLs := []string{"https://some.website.com"}
+//	patterns := []string{"website.com"}
+//
+//	dl, err := deadlinks.New(ctx, store, pinnedURLs, patterns, nil)
+//
+// `Update` is then used to crawl all links, starting with `pinnedURLs`:
+//
+//	err := dl.Update(ctx, time.Now())
+//
+// Finally, `GetByStatus` can be used to query all discovered resources based on
+// their current status. To retrieve all resources which have some error
+// (indicating a broken link):
+//
+//	erroredResources, err := miter.ToSlice(
+//		ctx, dl.GetByStatus(deadlinks.ResourceStatusError),
+//	)
+//
+// Note that `GetByStatus` returns a `miter.Iterator`, see its documentation for
+// more options on how to use it beyond `ToSlice`:
+//
+// https://godocs.io/code.betamike.com/mediocregopher/mediocre-go-lib/miter
+//
 // # Storage
 //
 // By default DeadLinks uses an in-memory SQLite database for tracking the
@@ -12,9 +56,12 @@
 //	})
 //	defer store.Close()
 //
-//	dl, err := deadlinks.New(
-//		ctx, store, pinnedURLs, patterns, nil,
-//	)
+//	dl, err := deadlinks.New(ctx, store, pinnedURLs, patterns, nil)
+//
+// # Further Customization
+//
+// Most functionality of DeadLinks can be extended or superceded by injecting
+// alternate interface implementations via the various Opts structs.
 package deadlinks
 
 import (
@@ -33,13 +80,13 @@ import (
 // Opts are optional fields which can be provided to New. A nil Opts is
 // equivalent to an empty one.
 type Opts struct {
-	Client Client // Defaults to `NewClient(nil)`
-	Parser Parser // Defaults to `NewParser()`
+	NewClient func() Client // Defaults to `func () Client { return NewClient(nil) }`
+	Parser    Parser        // Defaults to `NewParser()`
 
 	// Concurrency determines the maximum number of URLs which can be checked
 	// simultaneously.
 	//
-	// Default: `runtime.NumCPU()`
+	// Default: `runtime.NumCPU() / 2`
 	Concurrency int
 
 	// OnError, if set, will be called whenever DeadLinks encounters an error
@@ -49,7 +96,7 @@ type Opts struct {
 	// RequestTimeout determines how long a request for a resource can run
 	// before the resource is considered unavailable.
 	//
-	// Default: 10 * time.Second
+	// Default: 1 * time.Minute
 	RequestTimeout time.Duration
 }
 
@@ -58,8 +105,8 @@ func (o *Opts) withDefaults() *Opts {
 		o = new(Opts)
 	}
 
-	if o.Client == nil {
-		o.Client = NewClient(nil)
+	if o.NewClient == nil {
+		o.NewClient = func() Client { return NewClient(nil) }
 	}
 
 	if o.Parser == nil {
@@ -67,11 +114,11 @@ func (o *Opts) withDefaults() *Opts {
 	}
 
 	if o.Concurrency == 0 {
-		o.Concurrency = runtime.NumCPU()
+		o.Concurrency = runtime.NumCPU() / 2
 	}
 
 	if o.RequestTimeout == 0 {
-		o.RequestTimeout = 10 * time.Second
+		o.RequestTimeout = 1 * time.Minute
 	}
 
 	return o
@@ -87,6 +134,7 @@ type DeadLinks struct {
 	opts     Opts
 	store    Store
 	patterns []*regexp.Regexp
+	clients  []Client
 }
 
 // New initializes and returns a DeadLinks instance which will track the
@@ -129,6 +177,11 @@ func New(
 		patterns: patterns,
 	}
 
+	d.clients = make([]Client, d.opts.Concurrency)
+	for i := range d.clients {
+		d.clients[i] = d.opts.NewClient()
+	}
+
 	if err := d.store.SetPinned(ctx, pinnedURLs); err != nil {
 		return nil, fmt.Errorf("pinning URLs: %w", err)
 	}
@@ -155,11 +208,15 @@ func (d *DeadLinks) shouldFollowURL(url URL) bool {
 	return false
 }
 
-func (d *DeadLinks) getURL(ctx context.Context, url URL) ([]URL, error) {
+func (d *DeadLinks) getURL(
+	ctx context.Context, client Client, url URL,
+) (
+	[]URL, error,
+) {
 	ctx, cancel := context.WithTimeout(ctx, d.opts.RequestTimeout)
 	defer cancel()
 
-	mimeType, body, err := d.opts.Client.Get(ctx, url)
+	mimeType, body, err := client.Get(ctx, url)
 	if err != nil {
 		return nil, err
 	}
@@ -178,14 +235,16 @@ func (d *DeadLinks) getURL(ctx context.Context, url URL) ([]URL, error) {
 }
 
 // checkURL only returns an error if storing the results of the check fails.
-func (d *DeadLinks) checkURL(ctx context.Context, url URL) error {
+func (d *DeadLinks) checkURL(
+	ctx context.Context, client Client, url URL,
+) error {
 	var (
 		now      = time.Now()
 		status   = ResourceStatusOK
 		errorStr string
 	)
 
-	outgoingURLs, err := d.getURL(ctx, url)
+	outgoingURLs, err := d.getURL(ctx, client, url)
 	if err != nil {
 		status = ResourceStatusError
 		errorStr = err.Error()
@@ -219,14 +278,14 @@ func (d *DeadLinks) update(
 
 	wg.Add(d.opts.Concurrency)
 	for i := 0; i < d.opts.Concurrency; i++ {
-		go func() {
+		go func(client Client) {
 			defer wg.Done()
 			for url := range ch {
-				if err := d.checkURL(ctx, url); err != nil {
+				if err := d.checkURL(ctx, client, url); err != nil {
 					d.onError(ctx, fmt.Errorf("checking url %q: %w", url, err))
 				}
 			}
-		}()
+		}(d.clients[i])
 	}
 
 	var (

flake.nix

@@ -35,7 +35,6 @@
               pkgs.go
               pkgs.gotools
               pkgs.golangci-lint
-              pkgs.sqlite
             ];
           };
         });

resource.go

@@ -32,14 +32,15 @@ func (ds ResourceStatus) String() string {
 // uniquely identified by a URL.
 type Resource struct {
 	URL         URL
-	Status      ResourceStatus
-	Pinned      bool
-	LastChecked time.Time
+	Status      ResourceStatus `yaml:"-"`
+	Pinned      bool           `yaml:"-"`
+	LastChecked time.Time      `yaml:"last_checked"`
 
 	// only set if Status == ResourceStatusError
-	ErrorString string
+	ErrorString string `yaml:"error"`
 
 	// Indicate the URLs of resources which link to/are linked from this
 	// resource.
-	IncomingLinkURLs, OutgoingLinkURLs []URL
+	IncomingLinkURLs []URL `yaml:"incoming_links"`
+	OutgoingLinkURLs []URL `yaml:"outgoing_links"`
 }