summaryrefslogtreecommitdiff
path: root/cmd/gen-posix
diff options
context:
space:
mode:
Diffstat (limited to 'cmd/gen-posix')
-rw-r--r--cmd/gen-posix/data.go211
-rw-r--r--cmd/gen-posix/http_hacks.go156
-rw-r--r--cmd/gen-posix/main.go214
-rw-r--r--cmd/gen-posix/types.go26
4 files changed, 607 insertions, 0 deletions
diff --git a/cmd/gen-posix/data.go b/cmd/gen-posix/data.go
new file mode 100644
index 0000000..165ecbd
--- /dev/null
+++ b/cmd/gen-posix/data.go
@@ -0,0 +1,211 @@
+package main
+
+import (
+ "fmt"
+ "os"
+ "os/exec"
+ "regexp"
+ "strings"
+
+ "git.lukeshu.com/www/lib/httpcache"
+)
+
+var IEEESA = Vendor{
+ Name: "IEEE-SA",
+ GetURL: func(id string) string { return fmt.Sprintf("http://standards.ieee.org/findstds/standard/%s.html", id) },
+ GetName: func(id string, url string) string {
+ html, err := httpcache.Get(url, nil)
+ if err != nil {
+ panic(fmt.Errorf("URL=%q: %v", url, err))
+ }
+ cmd := exec.Command("nokogiri", "-e", `puts $_.css("meta[name=\"des\"], meta[name=\"designation\"]").first["content"]`)
+ cmd.Stderr = os.Stderr
+ cmd.Stdin = strings.NewReader(html)
+ d, err := cmd.Output()
+ if err != nil {
+ panic(fmt.Errorf("URL=%q: %v", url, err))
+ }
+ return strings.TrimSuffix(string(d), "\n")
+ },
+}
+
+var reIEEE = regexp.MustCompile(`standardNumber":"([^"]*)"`)
+
+var IEEEXplore = Vendor{
+ Name: "IEEE Xplore",
+ GetURL: func(id string) string { return fmt.Sprintf("http://ieeexplore.ieee.org/servlet/opac?punumber=%s", id) },
+ GetName: func(id string, url string) string {
+ if strings.HasSuffix(url, "ERROR") {
+ return "ERROR"
+ }
+ html, err := httpcache.Get(url, nil)
+ if err != nil {
+ panic(fmt.Errorf("URL=%q: %v", url, err))
+ }
+ m := reIEEE.FindStringSubmatch(html)
+ if m == nil {
+ panic(fmt.Errorf("URL=%q did not contain expected JSON", url))
+ }
+ return m[1]
+ },
+}
+
+var TOG = Vendor{
+ Name: "The Open Group",
+ GetURL: func(id string) string { return fmt.Sprintf("https://www2.opengroup.org/ogsys/catalog/%s", id) },
+ GetName: func(id string, url string) string { return id },
+}
+
+var TOGOnline = Vendor{
+ Name: "online",
+ GetURL: func(id string) string { return fmt.Sprintf("http://pubs.opengroup.org/onlinepubs/%s/", id) },
+ GetName: func(id string, url string) string { return url },
+}
+
+var ISO = Vendor{
+ Name: "ISO",
+ GetURL: func(id string) string {
+ return fmt.Sprintf("http://www.iso.org/iso/home/store/catalogue_tc/catalogue_detail.htm?csnumber=%s", id)
+ },
+ GetName: func(id string, url string) string {
+ html, err := httpcache.Get(url, nil)
+ if err != nil {
+ panic(fmt.Errorf("URL=%q: %v", url, err))
+ }
+ cmd := exec.Command("nokogiri", "-e", `puts $_.css("[itemprop=\"name\"]").first.text`)
+ cmd.Stderr = os.Stderr
+ cmd.Stdin = strings.NewReader(html)
+ d, err := cmd.Output()
+ if err != nil {
+ panic(fmt.Errorf("URL=%q: %v", url, err))
+ }
+ return strings.TrimSuffix(string(d), "\n")
+ },
+}
+
+var Vendors = []Vendor{IEEESA, TOG, ISO}
+
+var Editions = []Edition{
+ {Name: "POSIX-2001 (Issue 6)", Docs: []Document{
+ {Vendor: IEEESA, Type: Full, ID: "1003.1-2001", Resellers: []Document{
+ {Vendor: IEEEXplore, Type: Full, ID: "7683"},
+ }},
+ }},
+ {Name: "----->XBD-2001", Docs: []Document{
+ {Vendor: TOG, Type: Full, ID: "C950"},
+ {Vendor: ISO, Type: Full, ID: "37312"},
+ }},
+ {Name: "----->XSH-2001", Docs: []Document{
+ {Vendor: TOG, Type: Full, ID: "C951"},
+ {Vendor: ISO, Type: Full, ID: "37313"},
+ }},
+ {Name: "----->XCU-2001", Docs: []Document{
+ {Vendor: TOG, Type: Full, ID: "C952"},
+ {Vendor: ISO, Type: Full, ID: "37314"},
+ }},
+ {Name: "----->XRAT-2001", Docs: []Document{
+ {Vendor: TOG, Type: Full, ID: "C953"},
+ {Vendor: ISO, Type: Full, ID: "37315"},
+ }},
+
+ {Name: "POSIX-2001, 2002 Edition", Docs: []Document{
+ {Vendor: IEEESA, Type: Patch, ID: "1003.1-2001-Cor_1-2002", Resellers: []Document{
+ {Vendor: IEEEXplore, Type: Patch, ID: "9507"},
+ }},
+ {Vendor: TOG, Type: Patch, ID: "U057", Resellers: []Document{
+ {Vendor: TOG, Type: Full, ID: "T031"},
+ }},
+ }},
+ {Name: "----->XBD-2001, 2002 Edition", Docs: []Document{
+ {Vendor: TOG, Type: Full, ID: "C031"},
+ {Vendor: ISO, Type: Full, ID: "38789", Resellers: []Document{
+ {Vendor: IEEESA, Type: Full, ID: "9945-1-2003"},
+ }},
+ }},
+ {Name: "----->XSH-2001, 2002 Edition", Docs: []Document{
+ {Vendor: TOG, Type: Full, ID: "C032"},
+ {Vendor: ISO, Type: Full, ID: "38790", Resellers: []Document{
+ {Vendor: IEEESA, Type: Full, ID: "9945-2-2003"},
+ }},
+ }},
+ {Name: "----->XCU-2001, 2002 Edition", Docs: []Document{
+ {Vendor: TOG, Type: Full, ID: "C033"},
+ {Vendor: ISO, Type: Full, ID: "38791", Resellers: []Document{
+ {Vendor: IEEESA, Type: Full, ID: "9945-3-2003"},
+ }},
+ }},
+ {Name: "----->XRAT-2001, 2002 Edition", Docs: []Document{
+ {Vendor: TOG, Type: Full, ID: "C034"},
+ {Vendor: ISO, Type: Full, ID: "38792", Resellers: []Document{
+ {Vendor: IEEESA, Type: Full, ID: "9945-4-2003"},
+ }},
+ }},
+
+ {Name: "POSIX-2001, 2004 Edition", Docs: []Document{
+ {Vendor: IEEESA, Type: Patch, ID: "1003.1-2001-Cor_2-2004", Resellers: []Document{
+ {Vendor: IEEEXplore, Type: Patch, ID: "9022"},
+ {Vendor: IEEEXplore, Type: Full, ID: "9156"},
+ }},
+ {Vendor: TOG, Type: Patch, ID: "U059", Resellers: []Document{
+ {Vendor: TOG, Type: Full, ID: "T041"},
+ {Vendor: TOGOnline, Type: Full, ID: "009695399"},
+ }},
+ }},
+ {Name: "----->XBD-2001, 2004 Edition", Docs: []Document{
+ {Vendor: TOG, Type: Full, ID: "C046"},
+ {Vendor: ISO, Type: Patch, ID: "40687"},
+ }},
+ {Name: "----->XSH-2001, 2004 Edition", Docs: []Document{
+ {Vendor: TOG, Type: Full, ID: "C047"},
+ {Vendor: ISO, Type: Patch, ID: "40688"},
+ }},
+ {Name: "----->XCU-2001, 2004 Edition", Docs: []Document{
+ {Vendor: TOG, Type: Full, ID: "C048"},
+ {Vendor: ISO, Type: Patch, ID: "40690"},
+ }},
+ {Name: "----->XRAT-2001, 2004 Edition", Docs: []Document{
+ {Vendor: TOG, Type: Full, ID: "C049"},
+ {Vendor: ISO, Type: Patch, ID: "40691"},
+ }},
+
+ {Name: "POSIX-2008 (Issue 7)", Docs: []Document{
+ {Vendor: TOG, Type: Full, ID: "C082", Resellers: []Document{
+ {Vendor: TOGOnline, Type: Full, ID: "9699919799.2008edition"},
+ }},
+
+ {Vendor: IEEESA, Type: Full, ID: "1003.1-2008", Resellers: []Document{
+ {Vendor: IEEEXplore, Type: Full, ID: "4694974"},
+ }},
+
+ {Vendor: ISO, Type: Full, ID: "50516", Resellers: []Document{
+ {Vendor: IEEESA, Type: Full, ID: "9945-2009"},
+ {Vendor: IEEEXplore, Type: Full, ID: "5393777"},
+ }},
+ }},
+ {Name: "POSIX-2008, 2013 Edition", Docs: []Document{
+ {Vendor: TOG, Type: Patch, ID: "U130", Resellers: []Document{
+ {Vendor: TOG, Type: Full, ID: "C138"},
+ {Vendor: TOGOnline, Type: Full, ID: "9699919799.2013edition"},
+ }},
+
+ {Vendor: IEEESA, Type: Patch, ID: "1003.1-2008-Cor_1-2013", Resellers: []Document{
+ {Vendor: IEEEXplore, Type: Patch, ID: "6482152"},
+ {Vendor: IEEEXplore, Type: Full, ID: "6506089"},
+ }},
+
+ {Vendor: ISO, Type: Patch, ID: "62005"},
+ }},
+ {Name: "POSIX-2008, 2016 Edition", Docs: []Document{
+ {Vendor: TOG, Type: Patch, ID: "U160", Resellers: []Document{
+ {Vendor: TOG, Type: Full, ID: "C165"},
+ {Vendor: TOGOnline, Type: Full, ID: "9699919799.2016edition"},
+ }},
+
+ {Vendor: IEEESA, Type: Patch, ID: "1003.1-2008-Cor_2-2016", Resellers: []Document{
+ {Vendor: IEEEXplore, Type: Patch, ID: "7542096"},
+ {Vendor: IEEEXplore, Type: Full, ID: "7582336"},
+ }},
+ }},
+}
+
+// SUSv2 http://pubs.opengroup.org/onlinepubs/007908799/
diff --git a/cmd/gen-posix/http_hacks.go b/cmd/gen-posix/http_hacks.go
new file mode 100644
index 0000000..16b8a8d
--- /dev/null
+++ b/cmd/gen-posix/http_hacks.go
@@ -0,0 +1,156 @@
+package main
+
+import (
+ "bufio"
+ "bytes"
+ "errors"
+ "fmt"
+ "io"
+ "net/http"
+ "os"
+ "os/exec"
+ "strings"
+
+ "git.lukeshu.com/www/lib/httpcache"
+)
+
+func _checkURL(url string) (string, error) {
+ switch {
+ case strings.HasPrefix(url, "https://web.archive.org/"):
+ _, err := httpcache.Get(url, nil)
+ return url, err
+ case strings.HasPrefix(url, "https://www2.opengroup.org/ogsys/catalog/"):
+ _, err := httpcache.Get(url, nil)
+ if err == nil {
+ return url, nil
+ }
+ if !errors.Is(err, os.ErrNotExist) { // don't hide non-404 errors
+ return "", err
+ }
+ suffix := strings.TrimPrefix(url, "https://www2.opengroup.org/ogsys/catalog/")
+ url2 := "https://publications.opengroup.org/" + strings.ToLower(suffix)
+ _, err = httpcache.Get(url2, nil)
+ if err == nil {
+ return url2, nil
+ }
+ if !errors.Is(err, os.ErrNotExist) { // don't hide non-404 errors
+ return "", err
+ }
+ url3, err := _checkURL("https://web.archive.org/web/20170102/" + url)
+ if err == nil {
+ return url3, nil
+ }
+ return url+"#ERROR", nil
+ case url == "http://ieeexplore.ieee.org/servlet/opac?punumber=7394900":
+ return url+"#ERROR", nil
+ default:
+ _, err := httpcache.Get(url, nil)
+ if err != nil && errors.Is(err, os.ErrNotExist) {
+ return _checkURL("https://web.archive.org/web/20170102/" + url)
+ }
+ return url, err
+ }
+}
+
+func checkURL(url string) string {
+ url2, err := _checkURL(url)
+ if err != nil {
+ panic(fmt.Errorf("URL=%q: %v", url, err))
+ }
+ return url2
+}
+
+func nokogiriIgnoreFailure(htmlBytes []byte, expr string) string {
+ cmd := exec.Command("nokogiri", "-e", "puts "+expr)
+ cmd.Stderr = io.Discard
+ cmd.Stdin = bytes.NewReader(htmlBytes)
+ outBytes, _ := cmd.Output()
+ return strings.TrimSpace(string(outBytes))
+}
+
+func mockRedirect(url string) *http.Response {
+ resp, err := http.ReadResponse(bufio.NewReader(strings.NewReader(""+
+ "HTTP/1.1 302 Found\r\n"+
+ "Location: "+url+"\r\n"+
+ "\r\n")), nil)
+ if err != nil {
+ panic(err)
+ }
+ return resp
+}
+
+func mockForbidden() *http.Response {
+ resp, err := http.ReadResponse(bufio.NewReader(strings.NewReader(""+
+ "HTTP/1.1 403 Forbidden\r\n"+
+ "\r\n")), nil)
+ if err != nil {
+ panic(err)
+ }
+ return resp
+}
+
+func modifyResponse(url string, entry httpcache.CacheEntry, resp *http.Response) *http.Response {
+ switch {
+ case strings.HasPrefix(url, "https://web.archive.org/"):
+ htmlBytes, _ := io.ReadAll(resp.Body)
+ _ = resp.Body.Close()
+
+ // native Wayback Machine redirect
+ redirect := nokogiriIgnoreFailure(htmlBytes, `$_.css("p.impatient a").first["href"]`)
+ if strings.HasPrefix(redirect, "https://web.archive.org/web/") {
+ return mockRedirect(redirect)
+ }
+
+ // silly TOG SSO
+ if strings.Contains(url, "sso.opengroup.org") {
+ if bytes.Contains(htmlBytes, []byte("document.forms.postbinding.submit()")) {
+ redirect := nokogiriIgnoreFailure(htmlBytes, `$_.css("#postbinding").first["action"]`)
+ if redirect != "" {
+ return mockRedirect(redirect)
+ }
+ }
+ if bytes.Contains(htmlBytes, []byte("General Authorization Error")) {
+ return mockForbidden()
+ }
+ }
+
+ // We drained resp.Body, so re-create it.
+ resp, err := http.ReadResponse(bufio.NewReader(strings.NewReader(string(entry))), nil)
+ if err != nil {
+ panic(err)
+ }
+ return resp
+ default:
+ return resp
+ }
+}
+
+type mock404 struct {
+ Msg string
+}
+
+// Is implements the interface for [errors.Is].
+func (e *mock404) Is(target error) bool {
+ return target == os.ErrNotExist
+}
+
+// Error implements [error].
+func (e *mock404) Error() string {
+ return e.Msg
+}
+
+func checkRedirect(req *http.Request, via []*http.Request) error {
+ // net/http.defaultCheckRedirect
+ if len(via) >= 10 {
+ return errors.New("stopped after 10 redirects")
+ }
+
+ // detect redirects that should be 404s
+ oldURL := via[len(via)-1].URL
+ newURL := req.URL
+ if (newURL.Path == "/" || newURL.Path == "") && !(oldURL.Path == "/" || oldURL.Path == "") {
+ return &mock404{Msg: fmt.Sprintf("should have been a 404: %q redirected to %q", oldURL.String(), newURL.String())}
+ }
+
+ return nil
+}
diff --git a/cmd/gen-posix/main.go b/cmd/gen-posix/main.go
new file mode 100644
index 0000000..6da598b
--- /dev/null
+++ b/cmd/gen-posix/main.go
@@ -0,0 +1,214 @@
+package main
+
+import (
+ "bytes"
+ "fmt"
+ "html/template"
+ "os"
+
+ "git.lukeshu.com/www/lib/httpcache"
+)
+
+var urls = map[string]string{}
+var names = map[string]string{}
+
+func (doc Document) URL() string {
+ if doc.ID == "" {
+ return ""
+ }
+ key := doc.Vendor.Name + "\000" + doc.ID
+ if _, have := urls[key]; !have {
+ urls[key] = checkURL(doc.Vendor.GetURL(doc.ID))
+ }
+ return urls[key]
+}
+
+func (doc Document) Name() string {
+ if doc.ID == "" {
+ var names []string
+ for _, full := range doc.Fulls() {
+ names = append(names, full.Name())
+ }
+ switch len(names) {
+ case 0:
+ return "???"
+ case 1:
+ return names[0]
+ default:
+ // BUG(lukeshu): Handle un-IDed Documents with
+ // multiple IDed resellers.
+ panic("TODO")
+ }
+ }
+ key := doc.Vendor.Name + "\000" + doc.ID
+ if _, have := names[key]; !have {
+ names[key] = doc.Vendor.GetName(doc.ID, doc.URL())
+ }
+ return names[key]
+}
+
+func (doc Document) Fulls() []Document {
+ var ret []Document
+ for _, reseller := range doc.Resellers {
+ if doc.ID != "" && reseller.Vendor.Name == doc.Vendor.Name && reseller.ID == doc.ID {
+ continue
+ }
+ if reseller.Type == Full {
+ ret = append(ret, reseller)
+ }
+ }
+ return ret
+}
+
+func (doc Document) Patches() []Document {
+ var ret []Document
+ for _, reseller := range doc.Resellers {
+ if doc.ID != "" && reseller.Vendor.Name == doc.Vendor.Name && reseller.ID == doc.ID {
+ continue
+ }
+ if reseller.Type == Patch {
+ ret = append(ret, reseller)
+ }
+ }
+ return ret
+}
+
+func (doc Document) AsFull() *Document {
+ if doc.Vendor.Name == "" && doc.ID == "" {
+ return nil
+ }
+ ret := doc
+ ret.Resellers = ret.Fulls()
+ switch doc.Type {
+ case Full:
+ // Nothing to do
+ case Patch:
+ if len(ret.Resellers) == 0 {
+ return nil
+ }
+ ret.Type = Full
+ ret.ID = func() string {
+ var ids []string
+ for _, reseller := range ret.Resellers {
+ if reseller.Vendor.Name == doc.Vendor.Name {
+ ids = append(ids, reseller.ID)
+ }
+ }
+ switch len(ids) {
+ case 0:
+ return ""
+ case 1:
+ return ids[0]
+ default:
+ panic("wut")
+ }
+ }()
+ if ret.ID != "" {
+ ret.Resellers = ret.Fulls()
+ }
+ default:
+ panic("uhh")
+ }
+ return &ret
+}
+
+func (doc Document) AsPatch() *Document {
+ if doc.Vendor.Name == "" && doc.ID == "" {
+ return nil
+ }
+ ret := doc
+ switch doc.Type {
+ case Full:
+ return nil
+ case Patch:
+ ret.Resellers = doc.Patches()
+ default:
+ panic("no")
+ }
+ return &ret
+}
+
+func (ed Edition) DocsOrdered() []Document {
+ // This chould be O(n), but this niaeve implementation is
+ // O(n^2). It's OK, n is small.
+ s := make([]Document, len(Vendors))
+ for i, vnd := range Vendors {
+ for _, doc := range ed.Docs {
+ if doc.Vendor.Name == vnd.Name {
+ s[i] = doc
+ }
+ }
+ }
+ return s
+}
+
+var tmpl = `{{define "document"}}{{if .}}
+ {{if .URL}}<a href="{{.URL}}" title="{{.Name}}">{{.Name}}</a>{{else}}{{.Name}}{{end}}
+ {{range .Resellers}}
+ <a href="{{.URL}}" title="{{.Name}}">({{.Vendor.Name}})</a>
+ {{end}}
+{{end}}{{end}}
+<!DOCTYPE html>
+<html lang="en">
+ <head>
+ <meta charset="utf-8">
+ <title>POSIX Editions</title>
+ <style>
+ body { font-size: 8px; }
+ table { border-collapse: collapse; }
+ th, td { border: solid 1px black; }
+ tr:not(:first-child):not(:nth-child(2)) th {
+ font-family: monospace;
+ text-align: left;
+ }
+ </style>
+ </head>
+ <body>
+ <p>There's a typo in the "standardNumber" in IEEE
+ Xplore's records forfor the 2004 edition of 1003.1;
+ it says 2014 instead or 2004. The actual document
+ says 2004 though.</p>
+ <table>
+ <caption><p>POSIX: C & Shell (1997-present)</p></caption>
+ <tr><td rowspan=2></td>{{range .Vendors}}<th colspan=2>{{.Name}}</th>{{end}}</tr>
+ <tr>{{range .Vendors}}<th>Full</th><th>Patch</th>{{end}}</tr>
+ {{range .Editions}}<tr>
+ <th>{{.Name}}</th>
+ {{range .DocsOrdered}}
+ <td>{{template "document" .AsFull}}</td><td>{{template "document" .AsPatch}}</td>
+ {{end}}
+ </tr>{{end}}
+ </table>
+ </body>
+</html>
+`
+
+func mainWithError() error {
+ httpcache.UserAgent = "https://git.lukeshu.com/www/tree/cmd/gen-posix"
+ httpcache.ModifyResponse = modifyResponse
+ httpcache.CheckRedirect = checkRedirect
+
+ tmpl := template.Must(template.New("page").Parse(tmpl))
+
+ var out bytes.Buffer
+ if err := tmpl.Execute(&out, map[string]interface{}{
+ "Vendors": Vendors,
+ "Editions": Editions,
+ }); err != nil {
+ return err
+ }
+ if err := os.WriteFile("public/posix/index.new.html", out.Bytes(), 0666); err != nil {
+ return err
+ }
+ if err := os.Rename("public/posix/index.new.html", "public/posix/index.html"); err != nil {
+ return err
+ }
+ return nil
+}
+
+func main() {
+ if err := mainWithError(); err != nil {
+ fmt.Fprintf(os.Stderr, "%s: error: %v\n", os.Args[0], err)
+ os.Exit(1)
+ }
+}
diff --git a/cmd/gen-posix/types.go b/cmd/gen-posix/types.go
new file mode 100644
index 0000000..9bb4c2d
--- /dev/null
+++ b/cmd/gen-posix/types.go
@@ -0,0 +1,26 @@
+package main
+
+type Vendor struct {
+ Name string
+ GetURL func(id string) string
+ GetName func(id string, url string) string
+}
+
+type Type int
+
+const (
+ Full Type = 0
+ Patch Type = 1
+)
+
+type Edition struct {
+ Name string
+ Docs []Document
+}
+
+type Document struct {
+ Vendor Vendor
+ Type Type
+ ID string
+ Resellers []Document
+}
generated by cgit v1.2.3-54-g00ecf (git 2.45.2) at 2025-07-16 13:06:00 +0000